Therefore, our aim was to better define the pathophysiology of sodium retention by determining the following in patients with refractory ascites treated with TIPS: 1) changes in systemic and renal hemodynamics, renal sodium homeostasis, neurohumoral profile, and central blood volume; and 2) the relation among these variables.

Methods

Top Methods Results Discussion Author & Article Info References

Approval for the study was granted by the Human Experimentation Committee of the University of Toronto, and all patients gave informed consent.

Patients

Seven patients (5 men and 2 women) 38 to 75 years of age (mean age, 60 ±4 years) with biopsy-proven cirrhosis and refractory ascites were studied prospectively from November 1993 to June 1994. Ten patients were referred during that period for assessment of suitability for TIPS insertion, and 3 were rejected (1 was diagnosed with inoperable carcinoma of the rectum, 1 was diagnosed with hairy-cell leukemia, and one had a thrombosed portal vein on Doppler ultrasonography). The cause of cirrhosis was alcohol use in 3 patients and viral hepatitis B virus in 2 patients; cirrhosis was cryptogenic in 2 patients. Baseline clinical and biochemical data and severity of liver disease according to the Pugh classification [12] (Appendix Table 1) are shown in Tables 1 and 2. Five patients were Pugh class B (score, 7 to 9) and 2 were Pugh class C (score, 10). Patients with cirrhosis caused by alcohol use had abstained from alcohol for at least 6 months before being entered into the study. All patients had had ascites for more than 6 months before study entry. Refractoriness was defined as a prolonged history of ascites unresponsive to 400 mg of spironolactone or 30 mg of amiloride plus up to 120 mg of furosemide daily for 2 weeks.

Study Protocol

After discontinuing therapy with diuretics, all patients were admitted to the Clinical Investigation Unit of Toronto Hospital on a 22 mmol/d sodium, 1 L/d fluid, caffeine-free diet; this diet was maintained before and after TIPS insertion. During a 1-week stabilization period, patients were monitored daily for weight, serum electrolyte concentrations, and 24-hour urinary sodium excretion. None of the patients chosen for the study had renal disease, cardiovascular disease, or sepsis; all had a patent portal vein. Cardiologic assessment included chest radiography, electrocardiography, two-dimensional echocardiography, and examination by a cardiologist. Patients were excluded if they had primary renal disease that was ruled out by a normal serum creatinine level, normal urine on microscopy, and normal kidney size on ultrasonography; they were also excluded if they had spontaneous bacterial peritonitis that was ruled out by at least two negative ascitic-fluid cultures. Patients with other sources of sepsis, including dental sepsis, were excluded. Doppler ultrasonography was done to ascertain the patency of the portal vein.

On the morning before TIPS insertion, all patients had blood drawn for the measurement of baseline plasma norepinephrine levels, plasma renin activity, and aldosterone concentrations and for renal hemodynamic and sodium handling studies. Glomerular filtration rate and renal plasma flow were measured using inulin and para-aminohippurate clearances, respectively, and renal vascular resistance was calculated. Lithium clearance was used as an indicator of proximal tubular reabsorption of sodium [13]. On the afternoon before TIPS insertion, central blood volume was measured using radionuclide angiography [14].

Insertion was done on study day 8 while patients were under intravenous sedation in the angiography suite. Cardiac output, as measured by thermodilution, portal vein pressure, and free and wedged hepatic venous pressures, was measured before and 30 minutes after insertion, and the hepatic venous pressure gradient or the corrected sinusoidal pressure was calculated.

On the morning after TIPS insertion, Doppler ultrasonography was done to assess the patency of the TIPS; a shunt flow velocity of more than 100 cm/s indicated a widely patent shunt. Ultrasonography was followed by repeated renal hemodynamic studies and measurements of urinary sodium concentrations, hormonal levels, and central blood volume. In patients with a shunt flow velocity of no more than 100 cm/s, an urgent angiographic study of the TIPS was done. Acute thrombosis was lysed with urokinase, and Doppler ultrasonography was repeated. The patients were observed in the Clinical Investigation Unit for an additional 5 days after TIPS insertion; prophylactic lactulose was begun after TIPS insertion.

All patients were reviewed in an outpatient clinic 2 weeks after TIPS insertion. One month after insertion, they were rehospitalized for 1 week for repeated measurements of hormonal levels and central blood volume, for Doppler ultrasonography to assess shunt patency, and for renal hemodynamic studies. These tests were done after 5 days of stabilization in the hospital.

Techniques

Inulin, Para-aminohippurate, and Lithium Clearances

The techniques used to measure inulin and para-aminohippurate clearances [15] and lithium clearances [16] have been described previously. We used these clearances as indices of glomerular filtration rate, renal plasma flow, and proximal tubular sodium reabsorption, respectively.

Central Blood Volume Measurements

The use of radionuclide angiography to measure central blood volume has been described [14]. Quality assurance studies done in our Nuclear Cardiology Laboratory have established the standard error of the estimate of left ventricular ejection fraction calculation to be less than 2% using semi-automated techniques. The standard error of the estimate of ventricular volume calculation is less than 5 mL [17].

Insertion of TIPS

Under sterile technique and after the patient had received local anesthesia and moderate intravenous sedation, the left branch of the portal vein was punctured from the anterior abdominal approach using a 22-G needle and ultrasonographic guidance. A Mandril guidewire (Cook, Inc., Bloomington, Indiana) was advanced through the needle into the portal vein and positioned across the confluence of the left and right portal veins. Next, the right internal jugular vein was punctured at the mid-neck under local anesthesia. Using the Seldinger technique, a 10-F sheath was inserted. A double-lumen Swan-Ganz catheter was then introduced through the sheath; right heart pressures were measured and cardiac output was measured by thermodilution. A 9-F Colapinto catheter (Cook, Inc.) was then inserted, and pressure measurements were obtained from the inferior vena cava and from the free and wedged hepatic veins. A transjugular liver biopsy needle (Cook, Inc.) was then inserted through the catheter, and the hepatic parenchyma was punctured under fluoroscopic guidance; the puncture was directed toward the Mandril wire. When blood was aspirated, a small injection of radiographic contrast medium was used to confirm cannulation of the portal vein. The Colapinto catheter was advanced over the needle into the portal vein, and then portal venography was done and portal pressure was measured. A 5-F angioplasty balloon (Cook, Inc.) was used to dilate the intrahepatic tract to 8 mm. A 68-mm Wallstent (Schneider, Inc., Richmond Hill, Ontario, Canada) was then advanced and deployed to lie in the parenchymal tract, extending at least 2 cm into the portal vein. If the stent did not completely traverse the tract to the hepatic vein, a second, overlapping stent was placed. The stent or stents were dilated to between 8 and 9 mm using a high-pressure angioplasty balloon. Portal venography and pressure measurements were repeated. If the portosystemic gradient exceeded 8 mm Hg, the stent or stents were re-dilated to 10 mm. After a resting period of 30 minutes, right heart pressures and cardiac output measurements were repeated. All catheters were then withdrawn and hemostasis was obtained. No anticoagulant was used during or after the procedure.

Laboratory Analysis

Serum and urinary sodium and lithium levels were measured using standard flame photometric techniques. Blood samples, to be used to measure plasma renin activity, aldosterone levels, and plasma norepinephrine concentrations, were collected in ice. The supernatant serum was stored at –70°C until analysis. Plasma renin activity and aldosterone levels were estimated using previously described techniques [18]. Plasma norepinephrine concentrations were measured using high-performance liquid chromatography as described by Eriksson and Persson [19] and Weicker and colleagues [20], with modifications. Plasma and urinary inulin and para-aminohippurate levels were estimated using chemical analyses according to the modified methods of Wasler and colleagues [21] and Brun [22].

Calculations

Inulin and para-aminohippurate clearances were corrected for body surface area and expressed per 1.73 m² body surface area. Renal vascular resistance (RVR) was calculated from the following formula: RVR = mean arterial pressure ÷ renal blood flow, where renal blood flow = renal plasma flow ÷ (1 –hematocrit). Proximal tubular reabsorption of sodium was calculated using lithium clearance and glomerular filtration rate; distal tubular reabsorption was calculated using inulin clearance, serum sodium concentrations, urinary sodium concentrations, and urinary volume [23].

Central blood volume, a measurement affected by body size, was corrected for body surface area using the patient's height and dry body weight. Dry body weight of patients with ascites was defined as body weight after complete paracentesis before the current hospitalization. Stroke volume, cardiac output, and systemic vascular resistance (SVR) were calculated from standard formulae [13]. Cardiac and central vascular blood volume were calculated as total central blood volume minus right and left pulmonary vascular volumes.

Corrected sinusoidal pressure = wedged hepatic venous pressure -free hepatic venous pressure.

Statistical Analysis

All results are expressed as means with 95% CIs. The results at day 1 and at 1 month after TIPS insertion have been tabulated as the means of the intra-patient differences from baseline, with the 95% CIs for the means of the intra-patient differences included. Differences over time were assessed using repeated-measures of analysis of variance. The Dunnett test was used to determine the statistical significance of differences between the baseline value for a variable and repeated measurements for the same variable. A P value of less than 0.05 was considered statistically significant.

Results

Top Methods Results Discussion Author & Article Info References

Clinical Course after TIPS Insertion

Insertion of TIPS was successful in all patients. The mean corrected sinusoidal pressure decreased from 18.2 mm Hg (95% CI, 13 to 29 mm Hg) to 7.7 mm Hg (CI, 5 to 10 mm Hg; P < 0.001) (Table 3). Procedure-related complications were noted in two patients. Doppler ultrasonography done 1 day after insertion showed that the shunt of one patient was occluded. This was confirmed by hepatic venography, and thrombolysis with 200 000 U of urokinase restored shunt patency. This patient did not receive further anticoagulation, and, at 1 month, his shunt remained patent. The hemoglobin level of another patient gradually decreased in the week after TIPS insertion, and paracentesis showed bloody ascites. Hepatic venography showed no evidence of a hepatic leak into the peritoneal cavity. No other sources of bleeding were obvious in this patient. She was given a blood transfusion and was observed; her hemoglobin level gradually stabilized in the second week after insertion.

All patients were followed for between 1 and 12 months after TIPS insertion. One patient refused to return after the 3-month follow-up assessment because of the distance involved in doing so. He died 9 months after insertion from variceal hemorrhage followed by the hepatorenal syndrome. The other six patients are alive and well; four are totally free of ascites and edema. Of these four, ascites disappeared at 1 month in two patients, at 3 months in one patient, and at 6 months in one patient. The other two patients had significant reduction in their ascites (at 2 months and 1 month after TIPS insertion, respectively); one has returned to work on a part-time basis.

Three patients had significant encephalopathy after TIPS insertion and required rehospitalization. In one patient, this was precipitated by the presence of bacterial peritonitis. Encephalopathy was controlled in all three patients by the use of increasing oral doses of lactulose, enemas, or both. All six patients have returned to the community and are leading relatively normal lives.

In all patients, the Pugh score was 9.67 (CI, 8 to 11) 1 day after TIPS insertion; this was significantly higher than the baseline score of 8.33 (CI, 7 to 10; P = 0.03). This difference was due to a significant increase in serum bilirubin levels at day 1 (43 µmol/L [CI, 17 to 69 µmol/L]) compared with baseline (19 µmol/L [CI, 10 to 29 µmol/L]; P = 0.03) and the development of encephalopathy in three patients. At 1 month, the Pugh score was still significantly higher at 10.00 (CI, 8 to 12; P = 0.05 compared with baseline). Serum bilirubin levels remained elevated at 36.6 µmol/L (CI, 21 to 56 µmol/L; P = 0.02 compared with baseline) and encephalopathy was still present in 2 patients. Serum albumin levels were 31.7 g/L (CI, 24 to 37 g/L) at baseline, 30.0 g/L (CI, 25 to 35 g/L) at day 1, and 31.8 g/L (CI, 21 to 36 g/L) at 1 month; and prothrombin time was 13.4 s (CI, 11.6 to 15.4 s) at baseline, 14.4 s (CI, 12.1 to 16.7 s) at day 1, and 14.6 s (CI, 12.4 to 17.2 s at 1 month); these values remained unchanged throughout the study period. Hemoglobin levels decreased from 118 g/L (CI, 101 to 134 g/L) at baseline to 106 g/L (CI, 88 to 119 g/L) 1 day after TIPS insertion and were still low at 104 g/L (CI, 94 to 116) at 1 month (Table 1).

Systemic and Renal Hemodynamics

Table 3 shows the changes that occurred in systemic hemodynamics after TIPS insertion. Cardiac output increased significantly and systemic vascular resistance decreased significantly immediately after insertion as measured by thermodilution. Hemodynamic measurements obtained using radionuclide angiography 1 day after insertion confirmed the measurements obtained with thermodilution. Follow-up using radionuclide angiography showed that these significant changes were more pronounced month 1 after insertion, by which time cardiac output had increased by 74% (CI, 10% to 145%; P = 0.03) and systemic vascular resistance had decreased by 32% (CI, 11% to 52%; P = 0.04) compared with baseline. Mean arterial pressure, however, remained unchanged throughout the study period (Table 1).

In contrast, the renal circulation did not significantly change during the study period (Table 2). Glomerular filtration rate decreased slightly; this decrease was accompanied by an increase in renal vascular resistance 1 day after TIPS insertion. These values, however, had returned to baseline levels by 1 month after insertion. Renal plasma flow did not change during the month.

Renal Sodium and Water Handling

Urinary volume during the 3-hour renal study period was 83 mL/h (CI, 47 to 108 mL/h) at baseline, 72 mL/h (CI, 40 to 96 mL/h) 1 day after TIPS insertion, and 74 mL/h (CI, 51 to 114 mL/h) at 1 month. Urinary sodium excretion was did not change from baseline to day 1 but had increased significantly by 1 month (P = 0.04) when measured either over a 24-hour period Table 2 or during the 3-hour renal study period (Figure 1). The increase in urinary sodium excretion at 1 month was due to a significant reduction in proximal tubular reabsorption of sodium (P = 0.01) (Figure 1). The distal tubular reabsorption of sodium was unchanged throughout the 1-month period.

View larger version (15K): [in this window] [in a new window]   Figure 1. Mean values for urinary sodium excretion and proximal tubular reabsorption of sodium before insertion of transjugular intrahepatic portosystemic stent shunt, 1 day after insertion and 1 month after insertion. P value is for the comparison with baseline values and is derived from analysis of variance. Bars indicate 95% CIs.

Hormonal Profile

At baseline, all mean neurohumoral factor levels were elevated. Mean plasma renin activity was 3.33 ng x L^–1 x s^–1 (CI, 0.90 to 5.30 ng x L^–1 x s^–1 [normal, 0.32 to 1.50 ng x L^–1 x s^–1 for a patient receiving a 22 mmol/d sodium diet who has been recumbent for > 6 hours]). The mean plasma aldosterone level was 2232 pmol/L (CI, 1626 to 2734 pmol/L [normal, 580 to 1240 pmol/L for a patient receiving a 22 mmol/d sodium diet]). The mean plasma norepinephrine level was 2.40 nmol/L (CI, 1.25 to 3.32 nmol/L [normal, <1.2 nmol/L for a patient receiving a 22 mmol/d sodium diet]). One day after TIPS insertion, plasma renin activity and plasma aldosterone levels were unchanged, but plasma norepinephrine levels had increased significantly to 3.18 nmol/L (CI, 1.48 to 4.37 nmol/L; P = 0.05 compared with baseline). When these variables were measured 1 month after insertion, both mean plasma renin activity (0.62 ng x L^–1 x s^–1 [CI, 0.05 to 1.35 ng x L^–1 x s^–1]; P = 0.02 compared with baseline) and mean plasma aldosterone levels (779 pmol/L [CI, 385 to 1754 pmol/L]; P = 0.02 compared with baseline) had decreased significantly. Plasma norepinephrine levels had also decreased (2.28 nmol/L [CI, 1.64 to 3.07 nmol/L]), but only to the elevated baseline levels seen before TIPS insertion (Figure 2).

View larger version (13K): [in this window] [in a new window]   Figure 2. Mean concentrations of plasma renin activity, aldosterone, and plasma norepinephrine before insertion of transjugular intrahepatic portosystemic shunt, 1 day after insertion and 1 month after insertion. P value is for comparison with baseline and is derived from analysis of variance. Bars indicate 95% Cls.

Central Blood Volume

Total central blood volume, pulmonary vascular volumes, and cardiac and central vascular volumes showed gradual increases relative to baseline from day 1 to 1 month after TIPS insertion. The differences, however, were not statistically significant (Table 3).

Discussion

Top Methods Results Discussion Author & Article Info References

Our results suggest that natriuresis associated with TIPS is delayed for as long as a month, that it is associated with significant reductions in antinatriuretic factors but not with changes in renal hemodynamics, and that it occurs in the presence of increased systemic vasodilatation.

In our study, TIPS insertion was done as a treatment for refractory that ascites in patients with cirrhosis. Apart from positive anecdotal reports about the procedure [1, 24, 25], the rationale for TIPS insertion came from two studies that showed that ascites did not occur in patients with a corrected sinusoidal pressure of less than 8 mm Hg [6, 7] and from another study in which corrected sinusoidal pressure correlated inversely with urinary sodium excretion [6]. Furthermore, two studies showed surgical portosystemic shunts to be effective in relieving cirrhotic ascites and reducing hyperreninemia in patients [26, 27], and another showed them to be effective in reducing renal sodium retention in dogs with cirrhosis [28]. Campbell and colleagues [29] showed a significant decrease in urinary sodium excretion with reversible partial hepatic vein constriction, which led to acute portal hypertension in the dog, whereas sodium excretion promptly returned to the baseline value on release of the hepatic vein ligation. Therefore, lowering sinusoidal portal hypertension with TIPS insertion should result in increased urinary sodium excretion.

Immediately after TIPS insertion, despite a significant reduction in sinusoidal portal pressure in our patients, natriuresis may have been prevented by an exacerbation of the hyperdynamic circulation and systemic vasodilatation as previously reported [8]. Sodium retention in cirrhosis has been ascribed to systemic vasodilatation [5]. In our patients, systemic hemodynamic changes that resulted in further systemic vasodilatation were evident 30 minutes after TIPS insertion with significant further increases in cardiac output and significant reductions in systemic vascular resistance. This was associated with an immediate further activation of the sympathetic nervous system as reflected by a further increase in plasma norepinephrine levels on day 1 after insertion. There was a tendency for renal vascular resistance to increase and the glomerular filtration rate to decrease. Together with the still significantly elevated plasma renin activity and aldosterone levels 1 day after TIPS insertion, this could have counteracted the natriuretic effects of reduced sinusoidal portal pressure.

The exacerbation of the hyperdynamic circulation immediately after TIPS insertion was evident both in our patients with ascites and in previously described patients without ascites [8]. Previous investigators found no correlation between the increase in cardiac index and the decrease in porto-atrial gradient, suggesting that the increase in cardiac output was not simply the result of an increase in cardiac preload. Indeed, our central blood volume measurements confirmed that volumes did not significantly increase in the pulmonary circulation, the cardiac chambers, or the great vessels immediately after TIPS insertion. Thus, the significant increase in systemic vasodilatation that occurred in the face of increased sympathetic nervous activity may have been due to increased quantities of vasodilators from the splanchnic circulation entering the systemic circulation through the shunt. Such vasodilators include nitric oxide [30] (which has a half-life of seconds [31]), glucagon [32], and bile acids [33, 34]. The hormonal profile on day 1 after TIPS insertion may be considered as compensatory for the significant vasodilatation that follows insertion.

Amelioration of ascites 1 month after insertion was associated with a significant increase in urinary sodium excretion. This was so despite further systemic vasodilatation, which was associated with an increase in total central blood volume, including cardiac and central vascular volumes. Plasma renin activity and plasma aldosterone levels had decreased to normal values; plasma norepinephrine levels had returned to the previous elevated baseline levels. In contrast, the renal circulation, the glomerular filtration rate, the serum sodium level, and, therefore, the filtered sodium load did not change. Thus, natriuresis 1 month after insertion may have been due to a decrease in angiotensin II levels consequent to a decrease in plasma renin activity, causing decreased proximal sodium reabsorption; this reabsorption has been shown to be exquisitely sensitive to minute doses of angiotensin II in such patients [35]. In addition, the normalization of serum aldosterone levels would lead to decreased distal sodium reabsorption. Alternatively, it is also possible that changes in portal pressure may directly affect renal function. Such a hepatorenal connection was established by Lang and coworkers [36], who, in studies of rats with cirrhosis, achieved a reduction in glomerular filtration rate by infusing glutamine into the superior mesenteric vein. This was abolished after transection of the spinal cord or renal denervation [36]. The extent of proximal renal tubular reabsorption of sodium has also been shown to correlate significantly with hepatic function both in patients with cirrhosis [16] and in animals [37].

Our results suggest that an interplay of portal and systemic hemodynamics and neurohumoral factors are important in the pathogenesis of sodium retention in patients with ascites. However, the beneficial effects of lowering portal pressure can be blunted by further systemic vasodilatation and increased plasma norepinephrine levels. A critical level of hepatic function has also been proposed in experimental animal models of cirrhosis: Below this level, sodium retention occurs [37]. Correlation between hepatic function and urinary sodium excretion has been shown in well-compensated patients with cirrhosis [16, 38]. The Pugh scores of this group of patients with cirrhosis and refractory ascites remained elevated or deteriorated further as natriuresis began; this was due to increasing bilirubin levels. This hyperbilirubinemia after TIPS insertion may have been caused by intravascular hemolysis secondary to mechanical damage to erythrocytes. This may have been related to turbulent flow in the metallic nonendothelialized stent [39-41] because there was a concomitant decrease in hemoglobin 1 day and 1 month after TIPS insertion. There was certainly no evidence that the increased sodium excretion was associated with a significant improvement in liver function.

In conclusion, TIPS insertion should not be done in any patients with refractory ascites without careful attention to cardiac and renal status because this procedure challenges cardiac function and may, initially, worsen renal function. This procedure is certainly not the panacea for all patients with ascites. It has complications, and increased sodium excretion is neither immediate nor does it completely return to normal. Therefore, continued sodium and fluid restriction after TIPS insertion is mandatory, and noncompliance is a contraindication. However, in suitable patients, TIPS is a safe and effective means of managing refractory ascites.