Chronic liver disease is often associated with changes in the cardiovascular system. Cardiac output may be increased and peripheral vascular resistance lowered, accompanied by the impairment of cardiovascular reflexes. Cardiovascular autonomic dysfunction might result in inadequate responses to stressful events.
Orthotopic liver transplantation (OLT) includes cross-clamping of the hepatic artery, portal vein, and inferior vena cava (IVC) in preparation for the anhepatic stage, which brings about such hemodynamic alterations as decreased venous return and cardiac output, together with increased systemic vascular resistance (SVR). During that stage the recipient's liver is removed.
Cardiovascular response to clamping of the IVC has been used as a marker to indicate whether venovenous bypass is required during the anhepatic stage to maintain renal function or to improve hemodynamic stability . Yet there has been no confirmation that venovenous bypass actually achieves these goals [2,3].
Additionally, significant cardiovascular deterioration may occur after reperfusion of the grafted liver and may result in postreperfusion syndrome (PRS). The etiology of this syndrome is not fully understood but has been attributed to a variety of factors, including metabolic acidosis, hyperkalemia, hypocalcemia, thromboembolism, and the release of vasoactive substances by the liver graft. PRS occurs in about 8%-30% of patients.
The object of the present study was to assess whether the hemodynamic response to clamping of the IVC during OLT may be of help in predicting hemodynamic behavior of the grafted liver during reperfusion and the likelihood that PRS will occur.
With the approval of our hospital's ethics committee, we performed a retrospective study on 94 consecutive adult patients who underwent OLT over a 30-mo period. Patients suffering from pretransplantation renal insufficiency and cases in which liver transplantation was performed to treat fulminating hepatic failure were excluded from the study because a venovenous bypass had been used during the anhepatic stage. These are the only two categories of patients in which venovenous bypasses are employed at our hospital.
Anesthesia was induced with thiopental and succinylcholine and maintained with diazepam (0.2 mg [center dot] kg-1 [center dot] h-1), fentanyl (10 micro g [center dot] kg-1 [center dot] h-1), pancuronium (0.05 mg [center dot] kg-1 [center dot] h-1), and isoflurane. Mechanical ventilation was employed, using a 40% oxygen/air mixture. End-tidal CO2 tension was maintained between 32 and 36 mm Hg. A radial artery and the pulmonary artery were catheterized and the following hemodynamic measurements taken: heart rate (HR), cardiac output (CO), mean arterial aveolar pressure (MAP), mean pulmonary artery pressure (MPAP), pulmonary wedge capillary pressure (PWCP) at end expiration, central venous pressure (CVP), cardiac index (CI), and systemic vascular resistance index (SVRI). Hemodynamic variable measurements were recorded at two separate times during the surgical procedure; 10 min before clamping of the IVC (CC - 10) and 10 min after clamping of the IVC (CC + 10).
After measuring the hemodynamic variables at CC - 10, PWCP was increased to more than 12 mm Hg by administration of crystalloids and/or blood products in order to normalize effective intravascular volume.
Before reperfusion, the grafted liver was washed with portal blood (approximately 300 mL) to rinse out the preservative solution, which was discharged through a tube inserted in the infrahepatic IVC. Before washing out the preservative solution, an amount of blood equal to the amount of fluid that was going to be lost was transfused. Wisconsin solution was used as the preservative in all the grafted livers.
Hyperventilation with O2 commenced immediately before reperfusion, and CaCl2 and bicarbonate were administered to prevent symptomatic hyperkalemia and acidosis, as called for by the prereperfusion analysis results, irrespective of the hemodynamic response to clamping of the IVC.
PRS was defined as a decrease in the MAP of more than 30% of the baseline value for more than 1 min during the first 5 min after reperfusion of the graft. Subjects were divided into two groups: patients who developed PRS (PRS group) and those who did not (non-PRS group).
The variables considered are expressed as percentages when measured qualitatively and as means (centering) with the standard deviation (spread) when measured quantitatively. The Kolmogorov-Smirnov test was used to determine whether the data were normally distributed. A paired t-test was employed to compare the measurements taken 10 min before and 10 min after clamping of the IVC (level of statistical significance P < 0.05). The following variable classifications were established to investigate which hemodynamic factors may contribute to the onset of PRS: a) preclamping variables (taken 10 min before clamping): HR, MAP, MPAP, CVP, PWCP, CI, SVRI; b) postclamping variables (taken 10 min after clamping): HR, MAP, MPAP, CVP, PWCP, CI, SVRI; and c) percentage variables (between the two measurement times): HR, MAP, MPAP, CVP, PWCP, CI, SVRI.
In order to preselect the variables for potential inclusion in the logistic regression model, univariate logistic regressions were performed for each of the above-mentioned parameters as independent variables and the occurrence of PRS as the dependent variable. Screening for P < 0.25 was performed as recommended by Hosmer and Lemeshow . After the preselection, multiple logistic regression was performed, exploring the different models using the ENTER method. The inclusion criterion in the maximum likelihood test was P < 0.05, and the exclusion criterion was P > 0.10. The convergence criteria applied were BCON and LCON (0.0001). Data processing and analysis were carried out using an IBM-compatible personal microcomputer with a database program and the SPSS/PC + V 5.0 (SPSS, Inc., Chicago, IL) statistical package.
According to the Kolmogorov-Smirnov test (P > 0.05), our data followed a normal distribution.
PRS occurred in 27 of the 94 patients in our series (28.7%). Table 1 summarizes the data for the patients in both the PRS and non-PRS groups and also lists the histological basis for the diagnosis of hepatobiliary disease. There were no significant differences between the two groups.
The mean duration of PRS was 4.90 +/- 6.12 min, with duration ranging from 2-30 min. The patients who developed PRS were treated effectively using a bolus of epinephrine (100-200 micro cg), except in five cases in which it was also necessary to administer an intravenous perfusion of dopamine (10 micro g [center dot] kg (-1) [center dot] min-1) and/or other alpha-adrenergic drugs.
(Table 2) presents the statistical description of the potential hemodynamic risk factors for the occurrence of PRS along with the statistical significance of the differences between the variable measurement values before and after clamping (paired t-test). The last column of Table 2 shows the significance for each parameter according to the univariate logistic regression; the values that met the screening criterion of P < 0.25 are marked.
(Table 3) summarizes the results of the multivariate analysis based on multiple logistic regression using the ENTER method. The variables that satisfied the univariate screening criterion (MAP CC - 10; MPAP CC - 10; PWCP CC - 10; CVP CC - 10; SVRI CC - 10; MAP CC + 10; CI CC + 10; HR CC + 10; % SVRI; % MPAP; % CI; % HR) were used as the independent variables, and occurrence or nonoccurrence of PRS was taken as the dependent variable. The output presented in the classification Table andthe statistics associated with the fit are also given.
A comparison of the hemodynamic variables before (CC - 10) and after (CC + 10) clamping of the IVC shows substantially decreased MAP, MPAP, PWCP, CVP, and CI values and increased SVRI values in both groups (Table 4). At CC + 10, the hemodynamic profile in the two groups was different. The SVRI value was higher in the non-PRS group than it was in the PRS group (P < 0.01). The CI value, on the other hand, was significantly lower in the non-PRS group (P < 0.05).
The percentage of change in the hemodynamic values recorded for the two groups also differed. The differences in the percentage of change in the CI, MPAP, and SVRI values were statistically significant between the two groups (P < 0.05). The non-PRS group displayed a decrease of 47.1% in the CI, compared with a 27.9% decrease in the PRS group. The SVRI increased by 49% in the PRS group as opposed to 85.7% in the non-PRS group.
To obtain reference values of use in predicting the occurrence of PRS, the percentage of change in the CI, SVRI, and MAP values were subdivided into two groups.
When the CI decreased by at least 50% of the preclamping value 5-10 min after clamping of the IVC (43 patients), the frequency of occurrence of PRS was 16.2% (n = 7); in contrast, when the decrease in CI was less than 50% of the preclamping value (51 patients), PRS occurred in 39.2% (n = 20; P < 0.05).
The decrease in the MAP was more than 30% at CC + 10 in only five patients in our series, three of whom developed PRS. Patients in whom the percentage of increase in the SVRI at CC + 10 was more than 50% of the value at CC - 10 were significantly less likely to develop PRS. A total of 45.9% of the patients (17 of 37) in whom the increase in the SVRI value at CC + 10 was less than or equal to 50% of the preclamping value developed PRS. When the converse was true (i.e., the increase was greater than 50% of the preclamping value), PRS occurred in only 17.5% of the patients (10 of 57) (P < 0.01). Accordingly, the increase percentage in the SVRI value displayed a sensitivity of 62.9% and a positive predictive value of 82.4%.
Circulation in patients suffering from end-stage liver disease has been described as hyperdynamic, with increased blood flow to the splanchnic organs, kidneys, and skeletal muscle. This results in a lower SVRI, splanchnic hyperemia, higher cardiac output, and decreased arterial blood pressure . The mechanisms responsible for these changes are not fully understood, but neural, physical, metabolic, and humoral factors have all been implicated .
During liver transplantation, the IVC is clamped immediately before the anhepatic stage, resulting in a 50%-60% decrease in venous return to the heart, thus producing substantially lower CI, MPAP, MAP, PWCP, and CVP values and increased HR, SVRI, and PVRI values [7,8]. These compensatory responses represent an attempt to restore MAP values within 5-10 min of clamping of the IVC and thus to attain an effective tissue perfusion pressure; in contrast, the CO value often remains at less than 50% of the preclamping value .
The frequency of occurrence of PRS in our series was similar to that reported by Aggarwal et al. , who used the same method of defining PRS as we have, although venous bypass processes were used in their study. However, it has not been conclusively shown that the use of venovenous bypasses would have decreased the frequency of occurrence of PRS in our patients and, indeed, Jugan et al.  recently found venovenous bypass to be ineffective in preventing PRS.
In an experimental model, the pressor response generated by the sympathetic adrenergic system was depressed in rats with portal hypertension . This reflex is mediated by baroreceptors located in the atrium, splanchnic vascular bed, and hepatic artery. Analysis of our results shows that this compensatory response was triggered in both groups of patients but that its intensity was different in the PRS and non-PRS groups. In the latter group, the response was greater at CC + 10, with a larger increase in vascular muscle tone reflected by an increase in the SVRI value, and, to a certain extent, this may be indicative of greater vascular adaptability. Others have confirmed that the peripheral vascular response to exogenous noradrenaline and angiotensin II is attenuated in cirrhotic patients [10-12].
On conclusion of the anhepatic stage, the supra and infrahepatic IVC and portal vein are unclamped, providing significant hemodynamic alterations as a result of a sharp upsurge in venous return coupled with the effects of hypothermia, hyperkalemia, acidosis, hyperosmolarity, hypocalcemia, and the release of vasoactive substances (air/particulate) during reperfusion of the graft . The result may sometimes be PRS. It has been suggested that the cause of the hypotension observed at the time of reperfusion appeared to be related chiefly to temporary vasodilatation that minimized the cardiodepressor effect, since they observed an increase in CO mediated by bioimpedance during reperfusion of the grafted liver.
We think that the hemodynamic response to clamping of the IVC, as measured by the percentage variation in the SVRI, could be a suitable variable for predicting hemodynamic stability on reperfusion. This was demonstrated in patients in whom higher increases in the SVRI value was strongly correlated with a lower rate of occurrence of PRS, because patients with the greatest vasoconstrictive response to compensate for the reduction in preload could be expected to respond to reperfusion of the grafted liver most efficiently and with less pronounced vasodilatation. This suggests that the baroreceptors in such patients remain more intact or that they are more sensitive to the adrenergic discharge triggered on reperfusion as a compensatory mechanism.
The decrease in preload in our patients was similar in the two groups, yet the increase in the intensity of postload differed, suggesting that the greater reduction in the CI in the non-PRS group was mainly caused by the greater increase in the SVRI in that group. Furthermore, we do not believe that patients in the non-PRS group exhibited lower cardiac contractility, since it would not appear feasible for cardiac contractility to have acted as a factor protecting them from PRS.
The response to clamping of the IVC during liver transplantation could be of great assistance to anesthesiologists in assessing the intactness of the cardiocirculatory system and its ability to cope with the substantial variations in the volume of intravascular blood that take place during the operation.
Ideally, high-risk patients (those with poor vasoconstriction) should be managed differently, namely, by seeking SVR mediators capable of reducing the onset/severity of the hypotension associated with PRS. However, the only measure shown to be truly effective in preventing PRS is action to counteract hypocalcemia and acidosis just before unclamping. We think that patients at risk for PRS should be treated preventively with vasoconstrictors before any substantial decrease in MAP takes place in the first five minutes after reperfusion.
We conclude that the hemodynamic response to clamping of the IVC in our series, as measured by the percentage of variation in the SVRI value, is useful in predicting the likelihood that PRS will occur in those patients in whom venovenous bypass is not employed. Accordingly, patients who are able to enhance their vascular muscle tone, and hence who are more effectively able to moderate the customary decrease in arterial pressure 10 minutes after clamping, could be expected to display a less intense vasodilatation response to reperfusion as a means of reducing the duration and extent of hypotension, thus reducing the frequency of occurrence of PRS.