Hyperbilirubinemia after TIPS creation has been attributed to hepatic impairment and TIPS-associated hemolytic anemia. Sanyal and colleagues [4] first described hemolysis after TIPS. In a subsequent prospective analysis [9], these investigators reported a 10% incidence of hemolytic anemia after TIPS. Erythrocyte damage caused by the metal struts of the stents used with TIPS superimposed on an erythrocyte membrane altered by cirrhosis were proposed as the mechanisms for TIPS-associated hemolysis [4, 9].

We describe severe hyperbilirubinemia after TIPS creation, report on the natural history of this hyperbilirubinemia, and identify laboratory and clinical predictors associated with the development of this condition.

Methods

Top Methods Results Discussion Author & Article Info References

From June 1990 to September 1996, 354 patients underwent TIPS creation at the University of California, San Francisco. Severe hyperbilirubinemia was defined as a change in total serum bilirubin level from less than 85.5 µmol/L before TIPS creation to greater than 171 µmol/L 1 month after TIPS creation. A prospectively collected clinical and laboratory database of all patients undergoing TIPS creation at our institution was reviewed to identify patients who met the definition of severe hyperbilirubinemia and controls. Patients were excluded from analysis if an obvious source for hyperbilirubinemia was identified or if insufficient data were available. Of the 23 patients who met the case definition, 1 had hepatic metastases, 1 had acetaminophen toxicity, and 2 had hepatic graft rejection. Thus, 19 patients with hyperbilirubinemia were included in the analysis. Of the 331 remaining patients, 73 had total serum bilirubin levels of 85.5 µmol/L or greater before TIPS creation, 13 had total serum bilirubin levels of 85.5 µmol/L or greater and 171 µmol/L or less 1 month after TIPS creation, 9 were 18 years of age or younger, 9 had previously undergone liver transplantation, 9 had had unsuccessful TIPS, and 5 had insufficient data. The remaining 213 patients formed the control group. The TIPS database was then reviewed for laboratory data and clinical outcomes.

Nine laboratory and clinical variables were evaluated by univariate logistic regression [10] as possible predictors of hyperbilirubinemia: pre-TIPS serum levels of aspartate aminotransferase, albumin, and creatinine; prothrombin time; age; Child-Pugh cirrhosis class C; preexisting portosystemic encephalopathy; indication for TIPS; and cause of liver disease. These features were also assessed by multivariate logistic regression to determine their independent associations with the development of severe hyperbilirubinemia after TIPS creation.

Results

Top Methods Results Discussion Author & Article Info References

The demographic and clinical characteristics and outcomes of patients with hyperbilirubinemia and controls are summarized in the (Table 1). The mean age, age distribution, and indications for TIPS were similar in the two groups.

Severe hyperbilirubinemia occurred in 5.4% (19 of 354) of all patients undergoing TIPS at our institution. Among these 19 patients, the mean total serum bilirubin level (±SD) increased from 47.9 ± 18.8 µmol/L before TIPS to a peak of 543.8 ± 287.3 µmol/L (range, 177.8 to 1026.0 µmol/L) within 1 month after TIPS creation (Figure 1, top left). A concomitant increase in the mean prothrombin time from 16.1 ± 1.9 seconds to 19.4 ± 5.6 seconds (range, 13.6 to 34.8 seconds) occurred within 1 month after TIPS creation (Figure 1, top right). An abrupt 4.5-fold or greater increase in the serum aspartate aminotransferase level (Figure 1, bottom left) was seen 1 day after TIPS creation in 7 of 19 patients. The total serum bilirubin level and the AST level began to increase simultaneously. However, the peak total serum bilirubin level lagged behind the peak AST level by days to weeks in all 7 patients.

View larger version (31K): [in this window] [in a new window]   Figure 1. Liver function and hematologic indices among 19 patients with hyperbilirubinemia 1 day before insertion of the transjugular intrahepatic portosystemic shunt (TIPS), 1 day after insertion, and within or at 30 days after insertion. Top left. Total serum bilirubin level. Top right. Prothrombin time. Bottom left. Aspartate aminotransferase (AST) level, expressed as fold-increase above the patient's pre-TIPS level. Bottom right. Serum hemoglobin level.

Evaluation for hemolysis included serial measurement of hemoglobin levels and fractionation of the total serum bilirubin level. In all patients, the mean hemoglobin level (100.3 ± 10.5 g/L) remained stable 1 week and 1 month after TIPS creation (Figure 1, bottom right); in 14 of 14 patients tested, indirect bilirubin was less than 80% of the total serum bilirubin level. The initial fluctuations in hemoglobin levels represent blood loss and subsequent transfusion among patients who underwent TIPS creation for refractory gastrointestinal bleeding (Figure 1, bottom right). In addition, the corrected reticulocyte count was 1% or less in 5 of 5 patients tested, and 3 of 3 peripheral blood smears available for review did not show hemolysis.

By 90 days after TIPS creation, 95% of patients with hyperbilirubinemia had died or had undergone liver transplantation compared with 17% of controls (P < 0.001). The 1 patient with hyperbilirubinemia who survived beyond 3 months died 8 months after TIPS creation. The leading causes of death among patients with hyperbilirubinemia and controls were similar and included multiorgan failure, hepatic failure, and sepsis. The indication for liver transplantation was progressive liver failure in all 6 patients with hyperbilirubinemia who underwent transplantation. In contrast, only 1 control underwent expedited liver transplantation because of rapid liver decompensation.

Univariate logistic regression revealed that the following pre-TIPS clinical features were associated with hyperbilirubinemia: a prothrombin time of 17 seconds or more (32% of patients with a prothrombin time 17 seconds compared with 6% of patients with a prothrombin time <17 seconds; P < 0.001), Child-Pugh cirrhosis class C (16% of patients with class C cirrhosis compared with 4% of patients with class A or B cirrhosis; P = 0.0046), preexisting portosystemic encephalopathy (14% of patients with encephalopathy compared with 5% of patients without encephalopathy; P = 0.025), and nonalcoholic causes of liver disease (12% of patients with alcoholic causes compared with 3% of patients without alcoholic causes; P = 0.029). Laboratory and clinical variables not associated with severe hyperbilirubinemia after TIPS creation included age; indication for TIPS; and pre-TIPS serum levels of aspartate aminotransferase, albumin, and creatinine.

A multivariate logistic regression model for predicting the development of hyperbilirubinemia included nonalcoholic causes of liver disease (odds ratio, 5.8 [95% CI, 1.5 to 22.2]), Child-Pugh class C (odds ratio, 3.0 [CI, 0.96 to 9.4]), and a prothrombin time of 17 seconds or more (odds ratio, 5.2 [CI, 1.4 to 20.0]). The addition of preexisting encephalopathy neither improved this model nor achieved statistical significance (P < 0.2).

Discussion

Top Methods Results Discussion Author & Article Info References

Development of severe hyperbilirubinemia after TIPS creation was associated with progressive hepatic deterioration that ultimately resulted in death or need for liver transplantation in all patients with hyperbilirubinemia. Patients who died had not been considered for transplantation because of recent substance abuse, medical noncompliance, adverse psychosocial circumstance, or severe comorbid illness. The indication for TIPS in most patients was life-threatening gastrointestinal bleeding. Two of the clinical and laboratory variables predictive of hyperbilirubinemia-Child-Pugh class C and a prothrombin time of 17 seconds or more-are markers of poor hepatic functional reserve. The association of hyperbilirubinemia with causes of liver disease other than alcohol is not unprecedented. A higher incidence of complications related to portosystemic shunting, such as encephalopathy, has been documented in nonalcoholic persons [5].

Because our study was retrospective and is therefore limited to the analysis of previously collected data, we cannot absolutely exclude hemolysis as a contributing factor in the development of hyperbilirubinemia in most patients. However, the stability of hemoglobin levels after TIPS creation; the relatively low percentage of indirect bilirubin values; and, most important, the natural history of patients with hyperbilirubinemia argue against hemolysis as a primary source of hyperbilirubinemia. Hepatic ischemia caused by reduced sinusoidal perfusion may have been an initiating factor for hepatic decompensation in the seven patients who developed a characteristic abrupt increase in aspartate aminotransferase level after TIPS creation. In 5 of these patients, a clinical course similar to that of subacute hepatic failure [11] ensued, with increases in aspartate aminotransferase levels ranging from 9- to 61-fold occurring immediately after TIPS creation.

Deterioration of liver function after portal diversion with portosystemic shunts is well documented in the surgical literature. Survival analyses based on the results of prospective, controlled trials of prophylactic and therapeutic surgical shunts have revealed excess postoperative mortality caused by liver failure [12-15]. Hepatic decompensation after surgical shunting has been attributed to portal diversion with failed hepatic artery compensatory dilatation [16]. This problem led to the development of surgical methods to preserve prograde portal blood flow, such as distal splenorenal and small-diameter H-graft shunts. Neither prograde blood flow nor postoperative survival, however, has been shown to uniformly improve with these methods [17].

Our findings are similar to those in the surgical literature and underscore the importance of assessing hepatic functional reserve before considering portosystemic shunt. The potential for shunt-related complications, such as liver failure, limits the applicability of TIPS. Clearly, the most judicious use of TIPS at this time is either as a bridge to liver transplantation or in patients with life-threatening bleeding in whom conventional therapy has failed. We strongly recommend a cautious approach to patients with advanced liver disease who are being considered for TIPS; severe hyperbilirubinemia after TIPS should prompt an expedited evaluation for liver transplantation.

Dr. Roberts: Division of Transplant Surgery, M-896, University of California, San Francisco, 505 Parnassus Avenue, Box 0780, San Francisco, CA 94143.

Ms. Gee and Dr. Bacchetti: Center for Knowledge Management, CL-113, University of California, San Francisco, 530 Parnassus Avenue, Box 0840, San Francisco, CA 94143.