A prognostic model for patients with end-stage liver disease

A cost-utility analysis of secondary prophylaxis for variceal hemorrhage

BACKGROUND

Secondary prophylaxis for esophageal variceal hemorrhage (VH) is recommended, but there has never been a cost-utility analysis of its implementation.

OBJECTIVE

The objective was to compare the cost utility of various strategies for the secondary prophylaxis of VH including (a) observation alone, (b) medical therapy (MED), (c) endoscopic band ligation (EBL), (d) endoscopic band ligation plus medical therapy (EBL + M), and (e) transjugular intrahepatic portosystemic shunt (TIPS), and to examine the effect of adherence on these strategies.

METHODS

A Markov model was developed for all five strategies, and included surveillance, risk of hepatic encephalopathy, complications, and nonadherence.

DATA SOURCES

Published literature and the Health Care Financing Administration.

TARGET POPULATION

People with cirrhosis and a history of controlled VH.

TIME HORIZON

Three years.

PERSPECTIVE

Third-party payer.

OUTCOME MEASURES

Incremental cost-effectiveness ratios for quality-adjusted life-years (QALYs) gained.

RESULTS OF BASE-CASE ANALYSIS

Combination EBL + M was the optimal strategy, dominating all other strategies including observation, meaning that it was more effective and less expensive than the others. In addition, EBL alone dominated observation and TIPS in terms of QALYs, and MED alone dominated the strategy of observation in terms of QALYs.

RESULTS OF SENSITIVITY ANALYSIS

Important variables affecting the optimal strategy were the odds ratio (OR) of VH with EBL compared to MED, the OR of VH with EBL + M compared to EBL, and patients' preferences regarding taking the medication as reflected in the associated toll exacted on the health state utility. Variations in these parameters within the range of clinical plausibility allowed EBL or MED to become the optimal strategy. TIPS was the optimal strategy only if adherence rates for all strategies were less than 12%. RESULTS OF MONTE CARLO ANALYSIS: Neither observation nor TIPS was ever the optimal strategy, and EBL + M was optimal in 62% of cases. If the variables identified in the sensitivity analysis were controlled, then EBL + M was optimal in 95% of cases.

CONCLUSIONS

TIPS should be reserved only for patients with very poor adherence. Otherwise, patients are best served by medications, EBL, or a combination of both, depending on the comparative rates of rebleeding with each and patients' preferences regarding medical therapy. The redundancy of combination band ligation plus medical therapy can improve outcomes, particularly in the setting of poor patient adherence.

Predictive factors for early mortality following liver transplantation

AIMS

To retrospectively review our liver transplant performance to identify factors that influenced early outcomes and to prospectively test their validity in predicting outcomes.

METHODS

Clinical records from 190 patients with liver transplants (LT; n = 200) performed between 1991 and 1997 were reviewed and the data evaluated by univariate and multivariate analyses regarding clinical outcome. The prognostic model thus obtained was prospectively evaluated in 55 patients undergoing transplant between 1999 and 2000.

RESULTS

Main indication for transplant was post-necrotic cirrhosis (61%), mostly HCV(+). The majority of patients were Child-Pugh C status (46%). Post-operative mortality at 3 months was 15.3%. Risk factors predicting death were: Child-Pugh C status (OR 1.3), pre-LT renal insufficiency (OR 5.8), malnutrition (OR 2.9) and technically complex surgery requiring cross-clamping with or without bypass (OR 4.9). None of the donor factors was significant. Prospectively applied to predict outcome in the 55 patients, the model had a sensitivity of 80% and a specificity of 88.8% with a higher-than-anticipated accuracy with a positive predictive value of 61.5% and a negative predictive value of 95.3%.

CONCLUSIONS

Pre-LT renal insufficiency is the most significant risk factor for early mortality and suggests that LT should be performed before evidence of irreversible renal insufficiency becomes manifest.

Relationship of the model for end-stage liver disease (MELD) scale to hepatic encephalopathy, as defined by electroencephalography and neuropsychometric testing, and ascites

OBJECTIVE

It has recently been suggested that the Model for End-Stage Liver Disease (MELD) is a better and a more objective predictor of mortality in patients with end-stage liver disease. The aim of our study was to determine the relationship of the MELD score to hepatic encephalopathy (HE), as determined by electroencephalography (EEG) and clinical and neuropsychometric examination, and ascites.

METHODS

A total of 66 patients underwent EEG, a neuropsychometric screening by Mini-Mental State Examination, Trails Making Tests, Rey-Osterreith Complex Figure, and Hopkins Verbal Learning Tests, and a clinical assessment for HE. The MELD score was calculated as previously described by using serum creatinine, bilirubin, and international normalized ratio. Subclinical HE was diagnosed if clinical examination did not detect HE but neuropsychometric tests and EEG were abnormal.

RESULTS

Sixteen patients had no HE, 28 had subclinical HE, and 22 had clinical HE. Age, sex, race, and cause of liver disease were similar in all three groups. Child-Turcotte-Pugh score was significantly higher in patients with clinical HE compared with the other two groups. There was only a modest correlation (r = 0.5) between Child-Turcotte-Pugh and the MELD scores. The distribution and mean MELD scores were similar in patients with or without HE as determined by clinical or neuropsychometric examination and EEG. Approximately 90% of patients with clinical HE or abnormal EEG and neuropsychometric tests had a MELD score less than 25. Similarly, the MELD score was not affected by the severity of ascites.

CONCLUSION

The MELD score does not correlate well with severity of HE or ascites. Patients with HE and ascites might not receive liver transplantation in a timely manner if MELD scores were to be used exclusively for organ allocation.

Mice expressing the human CYP7A1 gene in the mouse CYP7A1 knock-out background lack induction of CYP7A1 expression by cholesterol feeding and have increased hypercholesterolemia when fed a high fat diet

Cholesterol 7alpha-hydroxylase (CYP7A1) catalyzes the rate-limiting step in the pathway responsible for the formation of the majority of bile acids. Transcription of the gene is regulated by the size of the bile acid pool and dietary and hormonal factors. The farnesoid X receptor and the liver X receptor (LXR) are responsible for regulation by bile acids and cholesterol, respectively. To study the effects of dietary cholesterol and fat upon expression of the human CYP7A1 gene, mice were generated by crossing transgenic mice carrying the human CYP7A1 gene with mice that were homozygous knock-outs (CYP7A1(-/-)). The mice (mCYP7A1(-/-)/hCYP7A1) expressed the human gene at much higher levels than did the transgenics bred in the wild-type background. A diet containing 1% cholic acid reduced the expression of the human gene in mCYP7A1(-/-)/hCYP7A1 mice to undetectable levels. Cholestyramine (5%) increased the level of expression of the human gene and the mouse gene. Thus, farnesoid X receptor-mediated regulation was preserved. A diet containing 2% cholesterol increased expression of the mouse gene in wild-type mice, but it did not affect expression of the human gene in mCYP7A1(-/-)/hCYP7A1 mice. None of the diets altered the serum cholesterol or triglyceride levels in these mice; 1% cholic acid caused a redistribution of cholesterol from the high density lipoprotein to the low density lipoprotein density in the humanized mice but not in wild-type mice. A diet containing 30% saturated fat and 2% cholesterol caused a decrease in CYP7A1 levels in mCYP7A1(-/-)/hCYP7A1 mice. The serum cholesterol levels rose in all mice fed this diet. The increase was greater in the mCYP7A1(-/-)/hCYP7A1 mice. Together, these data suggest that the lack of an LXR element in the region from -56 to -49 of the human CYP7A1 promoter may account for some of the differences in response to diets between humans and rodents.

Determinants of early mortality in patients with decompensated chronic hepatitis B treated with antiviral therapy

BACKGROUND & AIMS

Chronic hepatitis B is a leading cause of death worldwide. To identify patients who might require urgent liver transplantation despite antiviral therapy, we investigated the determinants of early mortality in a large cohort of patients with decompensated chronic hepatitis B treated with lamivudine.

METHODS

One hundred fifty-four North American patients with decompensated chronic hepatitis B received lamivudine for a median of 16 months. Univariate and multivariate Cox regression modeling was used to develop a model of 6-month mortality.

RESULTS

A biphasic survival pattern was observed, with most deaths occurring within the first 6 months of treatment (25 of 32, 78%) because of complications of liver failure. The estimated actuarial 3-year survival of patients who survived at least 6 months was 88% on continued treatment. In multivariate modeling, elevated pretreatment serum bilirubin and creatinine levels as well as the presence of detectable hepatitis B virus (HBV) DNA (by the bDNA assay) pretreatment were significantly associated with 6-month mortality. An equation approximating the probability of early mortality was developed from these variables.

CONCLUSIONS

Our data demonstrate a distinct alteration in the slope of the survival curve after 6 months of lamivudine treatment for decompensated chronic hepatitis B. An equation consisting of 3 widely available pretreatment laboratory parameters was developed that can be used to predict the likelihood of early death in patients receiving lamivudine for decompensated chronic hepatitis B. These observations may help identify patients who can be stabilized with suppressive antiviral therapy vs. those who require urgent liver transplantation.

Predictors of mortality and resource utilization in cirrhotic patients admitted to the medical ICU

BACKGROUND AND OBJECTIVE

Cirrhotic patients admitted to the medical ICU (MICU) are associated with high mortality rates and high resource utilization. This study identifies specific predictors of increased mortality and resource utilization and uses them to develop and validate prognostic models in cirrhotic patients admitted to the MICU.

METHODS

Cirrhotic patients admitted to the MICU were identified from the Critical Care Section database (January 1993 to October 1998). Clinical data were extracted from chart review including hospital course variables, mortality, and length of stay (LOS). Total cost per case (TCPC) was obtained from the Transition System INC: Multivariate logistic and linear regression analyses identified the independent predictors of increased mortality and resource utilization used for model building (MB) and model validation (MV).

RESULTS

A total of 582 cases were randomized to the MB and MV groups. Each group contained 240 cases after exclusion criteria were applied. The MICU mortality rate was 36.6%, and the in-hospital mortality rate was 49.0%. Acute physiology, age, and chronic health evaluation (APACHE) III score (odds ratio [OR], 4.7; 95% confidence interval [CI], 2.70 to 8.16; p < 0.001), mechanical ventilation (OR, 4.57; 95% CI, 2.35 to 8.34); p < 0.001), and the use of pressors (OR, 7.57; 95% CI, 4.35 to 13.18; p < 0.001) were independent predictors of MICU mortality. APACHE III score (OR, 4.96; 95% CI, 2.97 to 8.29; p < 0.001), the use of pressors (OR, 6.55; 95% CI, 3.66 to 11.72; p < 0.001), and acute renal failure (ARF) (OR, 4.31; 95% CI, 2.41 to 7.71; p < 0.001) were independent predictors of in-hospital mortality. Increased LOS in the MICU was associated with mechanical ventilation, ARF, bronchoscopy, bacteremia, use of pressors, transjugular intrahepatic portosystemic shunt (TIPS), and never received cardiopulmonary resuscitation (CPR) (p < 0.005). Source of admission, platelet transfusion, bacteremia, pneumonia, and never received CPR were independently associated with increased total LOS (p < 0.001). Mechanical ventilation, platelet transfusion, bronchoscopy, TIPS, sepsis, and never received CPR were independent predictors of increased TCPC (p < 0.001).

CONCLUSION

Simple prognostic models for mortality and resource utilization have been developed for cirrhotic patients admitted to the MICU.

Outcomes of critically ill patients denied consideration for liver transplantation

Patients with advanced liver disease (ALD) leading to admission to the ICU are often evaluated for possible orthotopic liver transplantation (OLT). Those deemed ineligible for listing for OLT must be managed by medical therapy. The number of patients not eligible for OLT listing will likely increase given the current organ transplant shortage. We performed a retrospective multivariate analysis of mortality predictors for patients denied listing for OLT. One hundred and eighty-three patients denied OLT listing were evaluated over a 3(1)/(2)-yr period, beginning in 1994. Overall mortality was 56% for those not listed versus 12% for those listed for OLT (p < 0.001). Independent predictors of increased mortality among those not listed were APACHE II score (p = 0.001; OR 1.11), sepsis (p = 0.04; OR 2.41), and the need for mechanical ventilation (p = 0.001; OR 3.71). Gastrointestinal (GI) bleeding was associated with decreased mortality (p = 0.02; OR 0.44). We conclude that critically ill patients with ALD denied OLT listing have substantially higher mortality than those listed for OLT. APACHE II score, sepsis, and the need for mechanical ventilation predict increased mortality in this group. Conversely, GI bleeding predicts decreased mortality; therefore, aggressive resuscitative measures seem merited in these patients.

A model to predict poor survival in patients undergoing transjugular intrahepatic portosystemic shunts

Transjugular intrahepatic portosystemic shunts (TIPS) may worsen liver function and decrease survival in some patients. The Child-Pugh classification has several drawbacks when used to determine survival in such patients. The survival of 231 patients at 4 medical centers within the United States who underwent elective TIPS was studied to develop statistical models to (1) predict patient survival and (2) identify those patients whose liver-related mortality post-TIPS would be 3 months or less. Among these elective TIPS patients, 173 had the procedure for prevention of variceal rebleeding and 58 for treatment of refractory ascites. Death related to liver disease occurred in 110 patients, 70 within 3 months. Cox proportional-hazards regression identified serum concentrations of bilirubin and creatinine, international normalized ratio for prothrombin time (INR), and the cause of the underlying liver disease as predictors of survival in patients undergoing elective TIPS, either for prevention of variceal rebleeding or for treatment of refractory ascites. These variables can be used to calculate a risk score (R) for patients undergoing elective TIPS. Patients with R > 1.8 had a median survival of 3 months or less. This model was superior to both the Child-Pugh classification, as well as the Child-Pugh score, in predicting survival. Using logistic regression and the same variables, we also developed a nomogram that indicates which patients survive less than 3 months. Finally, the model was validated among an independent set of 71 patients from the Netherlands. This Mayo TIPS model may predict early death following elective TIPS for either prevention of variceal rebleeding or for treatment of refractory ascites.

Deoxycholic acid treatment in patients with cholesterol gallstones: failure to detect a suppression of cholesterol 7alpha-hydroxylase activity

Hillebrant C-G, Nyberg B, Angelin B, Axelson M, Björkhem I, Rudling M, Einarsson C (Huddinge University Hospital and Karolinska Hospital, Karolinska Institute, Stockholm, Sweden). Deoxycholic acid treatment in patients with cholesterol gallstones: failure to detect a suppression of cholesterol 7alpha-hydroxylase activity. J Intern Med 1999; 246: 399-407.

OBJECTIVES

Based on animal studies, hydrophobic bile acids have been postulated to be particularly strong inhibitors of bile acid synthesis. The present study was undertaken to characterize in humans the effects of one of the most hydrophobic of the common bile acids, deoxycholic acid (DCA), on the transcriptional regulation and activity of the cholesterol 7alpha-hydroxylase, on hepatic cholesterol metabolism and on biliary lipid metabolism and plasma lipids. DESIGN, SUBJECTS AND SETTINGS: Thirteen patients with cholesterol gallstone disease were treated with DCA (750 mg day-1) for 3 weeks prior to cholecystectomy. Blood samples were collected before and during treatment. At operation, a liver biopsy and gallbladder bile were obtained. Twenty-eight untreated gallstone patients undergoing cholecystectomy served as controls. The study was carried out at a university hospital.

RESULTS

Deoxycholic acid comprised 72 +/- 6% (mean +/- SEM) of total biliary bile acids in DCA-treated patients (n = 8), and 21 +/- 2% in the controls (n = 16; P < 0.001). Cholesterol saturation of gallbladder bile averaged 102% in both treated (n = 7) and untreated (n = 16) patients. Cholesterol 7alpha-hydroxylase and HMG CoA reductase activities and mRNA levels were not different between DCA-treated and untreated gallstone patients. The LDL receptor mRNA levels were similar in both groups of patients. Plasma levels of total cholesterol were lowered by 10% upon DCA treatment (P < 0.05).

CONCLUSIONS

Treatment with DCA did not significantly affect mRNA levels and activity of hepatic cholesterol 7alpha-hydroxylase or HMG CoA reductase in patients with cholesterol gallstones. There was no effect on the saturation of gallbladder bile, Thus, the present study could not verify that the hydrophobicity of the bile acid pool is a major factor regulating human hepatic cholesterol 7alpha-hydroxylase activity.