High-density lipoprotein cholesterol as an indicator of liver function and prognosis in noncholestatic cirrhotics

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2005;3:286 –291

High-Density Lipoprotein Cholesterol as an Indicator of Liver Function and Prognosis in Noncholestatic Cirrhotics ADIL HABIB*,‡ ANASTASIOS A. MIHAS,*,‡ SOUHEIL G. ABOU–ASSI,*,‡ LESLIE M. WILLIAMS,‡ EDITH GAVIS,‡ W. MICHAEL PANDAK,*,‡ and DOUGLAS M. HEUMAN*,‡ *Department of Medicine, Virginia Commonwealth University Health Systems, Richmond, Virginia; and ‡Department of Medicine, Hunter Holmes McGuire Veterans Affairs Medical Center, Richmond, Virginia

Background & Aims: The liver plays a central role in production and degradation of lipoproteins. Declining lipoprotein cholesterol may reflect deteriorating liver function. Methods: We reviewed the records of 248 veterans with noncholestatic cirrhosis followed in our clinics or referred for liver transplantation between January 1, 1997 and October 31, 2002 (analysis period) and confirmed our findings prospectively in 165 noncholestatic cirrhotic veterans newly referred for liver transplantation between November 1, 2002 and May 1, 2004 (validation period). Results: In the analysis group, albumin, bilirubin, INR, and Model for End-Stage Liver Disease (MELD) score correlated strongly with high-density lipoprotein (HDL) cholesterol, weakly but significantly with total cholesterol and very-low-density lipoprotein cholesterol (VLDL), and poorly with low-density lipoprotein cholesterol (LDL). Transplant-free mortality at 90, 180, and 365 days was 17/201 (8.5%), 19/173 (11.0%), and 38/119 (31.9%), respectively. Death at all 3 time points was associated with significantly lower initial levels of HDL, VLDL, and total cholesterol, but not LDL cholesterol. Of the lipoproteins, HDL was the best predictor of survival at 180 and 365 days (concordance statistics .86ⴞ.05 and .78ⴞ.05, respectively). By multivariate logistic regression, HDL cholesterol and MELD score were independent predictors of survival at 6 and 12 months. By Cox regression, HDL cholesterol below 30 mg/dL was associated with 3.4-fold increase in the hazard ratio for cirrhotic death. In the validation period, HDL cholesterol was confirmed to be significantly associated with death or transplantation at 6 or 12 months. Conclusions: HDL cholesterol in noncholestatic cirrhotic patients is a liver function test and an indicator of prognosis.

he liver plays a central role in cholesterol homeostasis. Liver is quantitatively a major site of de novo cholesterol synthesis. Nascent high-density lipoproteins (HDLs) and very-low-density lipoproteins (VLDLs) are synthesized and secreted into the circulation by the liver, and mature lipoproteins such as low-density lipoproteins (LDLs), intermediate-density lipoproteins, chylomicron remnants, and

T

HDL are taken up by the liver in a regulated, receptordependent manner. Total plasma cholesterol level decreases with decrease in liver function. Previous studies have reached conflicting conclusions regarding the effects of liver disease on different lipoprotein cholesterol fractions. Some of the controversy reflects the fact that many studies of cholesterol in liver disease have focused on cholestatic disorders such as primary biliary cirrhosis, in which biliary cholesterol and phospholipids accumulate in plasma as lipoprotein X. In this study, we examined a large cirrhotic population with noncholestatic liver diseases to test the hypothesis that one or more plasma lipoprotein cholesterol fractions, as measured by simple standard clinical laboratory methods, may serve as a useful indicator of liver function and prognosis in hepatic cirrhosis.

Materials and Methods Study Design We conducted a survey of records of veterans with hepatic cirrhosis, with the diagnosis established by biopsy examination or by clinical and radiographic findings. Data collection was performed in compliance with federal regulations and with the oversight and approval of the McGuire Institutional Review Board. The study was divided into 2 phases: an analysis phase and a validation phase. Analysis was performed in patients enrolled before November 1, 2002. Two groups were included: (1) patients followed-up in the Hepatology and Liver Transplant clinics at Richmond’s Hunter Holmes McGuire Department of Veterans Affairs Medical Center; and (2) patients referred from other Department of Veterans Affairs centers throughout the United States for consideration of orthotopic liver transplantation. Initial data for 248 patients identified through November 1, 2002, were Abbreviations used in this paper: HDL, high-density lipoprotein; INR, international normalized ratio; LDL, low-density lipoprotein; MELD, Model for End-Stage Liver Disease; VLDL, very-low-density lipoprotein. © 2005 by the American Gastroenterological Association 1542-3565/05/$30.00 PII: 10.1053/S1542-3565(04)00622-6

March 2005

HDL IN CIRRHOSIS

Table 1. Lipoprotein Cholesterol and Liver Function Parameters (Mean ⫾ SD), Patients Evaluated Through November 1, 2002 (Analysis Group) Compared With Patients Enrolled After November 1, 2002 (Validation Group) Analysis group N Cholesterol (mg/dL) Total LDL HDL VLDL Albumin level (g/L) Bilirubin level (mg/dL) Creatinine level (mg/dL) INR AST/ALT ratio MELD score

Mean ⫾ SD

Validation group N

Mean ⫾ SD

248 145.7 ⫾ 41.9 165 142.3 ⫾ 48.0 248 86.2 ⫾ 31.6 165 80.9 ⫾ 37.7 248 37.0 ⫾ 18.7 165 41.8 ⫾ 17.8 248 22.7 ⫾ 13.2 165 20.9 ⫾ 12.5 200 2.92 ⫾ .61 163 2.97 ⫾ .58 204 2.40 ⫾ 1.76 165 2.60 ⫾ 2.68 208 1.08 ⫾ .58 165 1.13 ⫾ .57 188 1.55 ⫾ .37 147 1.49 ⫾ .38 203 1.59 ⫾ .56 157 1.56 ⫾ .52 187 16.4 ⫾ 7.2 147 15.1 ⫾ 5.4

AST, aspartate transaminase; ALT, alanine transaminase.

reported in abstract form on December 6, 2002.1 To validate these findings we prospectively collected lipoprotein data on 165 additional cirrhotic patients newly referred for consideration of liver transplantation between November 1, 2002, and May 1, 2004 (validation period). Outcomes (death, liver transplantation) were determined from 3 sources: (1) review of individual medical records, accessed from multiple Department of Veterans Affairs sites by using computerized remote data access, (2) query of the Department of Veterans Affairs Central Office Liver Transplant Program database, and (3) query of the Social Security Death Index. All lipoprotein panels were determined in Veterans Affairs hospital clinical laboratories. Fractionation used standard homogeneous laboratory methods in which total cholesterol and triglyceride levels were measured directly, HDL cholesterol level was determined by repeat cholesterol determination after precipitation or adsorption of non-HDL, the proportion of VLDL cholesterol was calculated from the triglyceride level using the Friedewald formula, and the remaining cholesterol was assigned to the LDL fraction. The most current available lipoprotein profile results (before transplantation) through August 2002 (analysis group) or February 2004 (validation group) were recorded. Where available, we also recorded concurrent levels of albumin, international normalized ratio (INR), bilirubin, creatinine, aspartate transaminase, and alanine transaminase, and calculated a Model for End-Stage Liver Disease (MELD) score. For purposes of this study, liver function test data were considered concurrent if obtained within 30 days before or after lipoprotein analysis. Other data recorded included demographic information (age, sex) and cause of liver disease. Patients receiving treatment with hypolipidemic drugs were excluded, as were patients with cholestatic liver diseases (primary biliary cirrhosis, primary sclerosing cholangitis, sarcoidosis, autoimmune hepatitis, and biliary ductal obstruction).

287

Statistical Analyses The relationship between lipoprotein cholesterol fractions and other liver function tests was assessed by using Pearson correlation. Survival was determined from the date of lipoprotein cholesterol determination. The univariate relationship of individual lipoprotein and liver function tests to mortality at 90, 180, and 365 days was assessed by analysis of variance (ANOVA). Independent predictors of survival were identified by bivariate logistic regression using the forward likelihood ratio method with inclusion and exclusion threshold probabilities of .05 and 0.1, respectively. The sensitivity and specificity of individual lipoproteins as prognostic tests of mortality at each time point were determined by examining receiver operating characteristic curves. The effect of lipoprotein cholesterol level on survival was assessed by Cox proportional hazards analysis. All analyses were performed using SPSS for Windows (version 11.0; SPSS, Chicago, IL). Except where stated otherwise, cases were censored (dropped from follow-up) at the time of transplantation. Therefore, the analyses only address the relationship of lipoprotein cholesterol to survival in the presence of cirrhosis. We did not attempt to determine whether pretransplant lipoprotein levels are associated with survival after liver transplantation.

Results Complete lipoprotein profile determinations (pretransplant) were available with follow-up evaluation through November 1, 2002, for 248 noncholestatic cirrhotic patients, who form the basis of our initial retrospective analysis; 106 of these patients died or underwent transplantation before November 1, 2002. For the period from November 1, 2002, through May 1, 2004, we confirmed our findings in 165 additional noncholestatic cirrhotic patients newly referred for consideration of liver transplantation for whom complete lipoprotein data were available. The combined groups thus included 413 patients. Consistent with the population served by the Department of Veterans Affairs, 98% were men. The major causes of cirrhosis were hepatitis B or C (63%) and/or alcohol abuse (66%); both alcohol-induced and viral hepatitis were deemed to have contributed significantly to the pathogenesis of cirrhosis in 37%. Other noncholestatic causes such as cryptogenic cirrhosis, nonalcoholic steatohepatitis, ␣-1-antitrypsin deficiency, and hemochromatosis accounted for only 8% of all cases. The distribution of lipoprotein cholesterol and liver tests in the analysis and validation groups is shown in Table 1. In the analysis group, values (mean ⫾ SD) of total, LDL, HDL, and VLDL cholesterol were 147 ⫾ 42, 87 ⫾ 32, 38 ⫾ 18, and 23 ⫾ 13, respectively. Values (mean ⫾ SD) of albumin, bilirubin, and INR in this population were 2.95 ⫾ .63, 2.39 ⫾ 1.65, and 1.55 ⫾ .34, respectively. The MELD score ranged from 6 to 40, with mean ⫾ SD of 16.4 ⫾ 7.2. Previous data indicate

288

HABIB ET AL

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 3, No. 3

Table 2. Pearson Correlation Coefficients (R) Between Lipoprotein Cholesterol Fractions and Liver Function Tests, Analysis Group Cholesterol (lipoprotein) Total LDL HDL VLDL

Correlation

Albumin (g/L)

Bilirubin (mg/dL)

Creatinine (mg/dL)

INR

AST/ALT ratio

MELD score

R Significance R Significance R Significance R Significance

.437 ⬍.001 .140 .049 .520 ⬍.001 .361 ⬍.001

⫺.186 .008 .04 .573 ⫺.361 ⬍.001 ⫺.204 .003

⫺.179 .01 ⫺.034 .622 ⫺.259 ⬍.001 ⫺.143 .04

⫺.371 ⬍.001 ⫺.106 .148 ⫺.445 ⬍.001 ⫺.342 ⬍.001

⫺.267 ⬍.001 ⫺.123 .08 ⫺.304 ⬍.001 ⫺.132 .06

⫺.264 ⬍.001 .015 .84 ⫺.441 ⬍.001 ⫺.298 ⬍.001

NOTE. Significance determinations are 2-tailed. AST, aspartate transaminase; ALT alanine transaminase.

that a MELD score of 16 is associated with a median transplant-free survival of 2–3 years; thus, our population clearly had a high proportion of patients with advanced liver disease. Consistent with this, the life-table mortality at 90, 180, and 365 days from lipoprotein profile determination (censored at transplantation) was 17 of 223 (7.6%), 20 of 192 (10.4%), and 39 of 125 (31.2%), respectively. With respect to lipoprotein levels and liver function tests, the analysis and validation groups were similar. The association of individual lipoprotein fractions with liver function tests in the analysis group was assessed by Pearson correlation, as shown in Table 2. Of the lipoproteins, HDL cholesterol level was found to correlate most strongly with concurrent values of albumin, INR, bilirubin, and MELD score (all P ⬍ .001). Weaker but highly significant correlations also were noted for each of these liver tests with total cholesterol and VLDL cholesterol. Importantly, LDL cholesterol level correlated only marginally

with albumin level and did not correlate significantly with any of the other liver function tests or with MELD score. These findings support the hypothesis that levels of HDL cholesterol and, to a lesser extent, total and VLDL cholesterol are indicators of liver function. By univariate ANOVA (Table 3), we found in the analysis group that cirrhotic patients dying within 90, 180, or 365 days, compared with transplant-free survivors at each time point, exhibited significantly lower initial values of total, HDL, and VLDL cholesterol, but similar LDL cholesterol levels. In these analyses, cases were censored at transplantation. The largest proportional difference and greatest statistical significance was noted for HDL cholesterol levels. Differences in HDL cholesterol levels accounted for most of the differences in total cholesterol noted between deaths and survivors at each time point (Figure 1). Similar results were noted when transplanted patients were included and death or transplantation were taken as the end points (data not shown).

Table 3. Association of Lipoproteins and Liver Tests (Mean ⫾ SD) With Life-Table Survival at 90, 180, or 365 Days, Analysis Group 90-day survival (mortality 17/223, 8%)

Total cholesterol (mg/dL) LDL cholesterol (mg/dL) HDL cholesterol (mg/dL) VLDL cholesterol (mg/dL) Albumin (g/L) Bilirubin (mg/dL) Creatinine (mg/dL) INR AST/ALT ratio MELD score

180-day survival (mortality 20/192, 11%)

365-day survival (mortality 39/125, 32%)

Dieda

Aliveb

P

Dieda

Aliveb

P

Dieda

Aliveb

P

109 ⫾ 48 81 ⫾ 35 21 ⫾ 17 16 ⫾ 7 2.3 ⫾ .4 4.0 ⫾ 2.5 1.5 ⫾ 1.0 1.8 ⫾ .4 1.67 ⫾ 9.72 23.6 ⫾ 7.7

151 ⫾ 45 88 ⫾ 31 39 ⫾ 18 24 ⫾ 14 3.0 ⫾ .6 2.2 ⫾ 1.5 1.0 ⫾ .3 1.5 ⫾ .3 1.56 ⫾ .52 15.0 ⫾ 6.2

⬍.001 .133 ⬍.001 .011 ⬍.001 ⬍.001 ⬍.001 ⬍.001 .284 ⬍.001

129 ⫾ 44 93 ⫾ 39 19 ⫾ 16 17 ⫾ 7 2.4 ⫾ .7 3.8 ⫾ 2.4 1.5 ⫾ .9 1.9 ⫾ .5 1.72 ⫾ .68 23.4 ⫾ 8.3

153 ⫾ 41 89 ⫾ 32 40 ⫾ 17 25 ⫾ 14 3.1 ⫾ .6 2.1 ⫾ 1.5 1.0 ⫾ .3 1.5 ⫾ .3 1.50 ⫾ .51 14.7 ⫾ 6.0

.016 .552 ⬍.001 .020 ⬍.001 ⬍.001 ⬍.001 ⬍.001 .071 ⬍.001

140 ⫾ 42 95 ⫾ 35 26 ⫾ 15 20 ⫾ 8 2.7 ⫾ .6 2.9 ⫾ 1.9 1.1 ⫾ .6 1.7 ⫾ .4 1.46 ⫾ .40 18.7 ⫾ 8.0

157 ⫾ 43 90 ⫾ 34 42 ⫾ 18 26 ⫾ 14 3.2 ⫾ 0.6 1.9 ⫾ 1.3 .9 ⫾ .3 1.5 ⫾ .3 1.58 ⫾ .51 14.2 ⫾ 6.0

.044 .484 ⬍.001 .022 ⬍.001 .004 .021 .004 .173 .003

NOTE. Data censored at transplantation. P ⫽ significance level by univariate ANOVA (2-tailed). AST, aspartate transaminase; ALT, alanine transaminase. aDied without undergoing transplantation. bLiving with cirrhosis, not transplanted.

March 2005

HDL IN CIRRHOSIS

90 days

180 days

365 days

Cholesterol (mg/dl)

120 100 80 60

**

*

HDL

VLDL

**

*

HDL

VLDL

**

*

HDL

VLDL

40 20 0

LDL

Died Survived

LDL

LDL

* = p < 0.05 ** = p < 0.001

Figure 1. Association of lipoprotein cholesterol with mortality in noncholestatic cirrhotic patients. Analysis group: life-table mortality at 90, 180, or 365 days (censored at transplantation). Bars display means ⫾ 2 SEM. *P ⬍ .05; **P ⬍ .001. , Died; □, survived.

The independent significance of the various liver function tests and lipoproteins as predictors of survival at 90, 180, and 365 days was assessed in the analysis group using bivariate logistic regression (Table 4). Variables included in the analysis were total, HDL, LDL, and VLDL cholesterol levels, and MELD score. We found that MELD score was the only independent predictor of survival at 90 days. However, at 180 days HDL cholesterol level and MELD score both were independent predictors, and for 1-year survival the HDL cholesterol level was the only independent predictor identified. The sensitivity and specificity of the various lipoprotein fractions as predictors of mortality in the analysis group were assessed using receiver operating characteristic curves as shown in Table 5. The area under the receiver operating characteristic curve, or concordance statistic, is a measure of the test’s predictive accuracy, with 1 indicative of an ideal test and .5 indicating no predictive value. As shown, HDL level was a highly significant predictor of mortality at 90, 180, and 365 days, with concordance statistics (mean ⫾ SEM) of .86 ⫾ .06, .86 ⫾ .06, and .79 ⫾ .07, respectively. Total and VLDL cholesterol levels were weaker predictors, whereas LDL level was not predictive of mortality at any of the 3 time points. Predictive value of HDL level was comparable with that of the MELD score at 90, 180, and 365 days (MELD concordance statistics .81 ⫾ .06, .79 ⫾ .07, and .69 ⫾ .06, respectively). The use of low HDL cholesterol level as a predictor of mortality was confirmed in the validation period. As in the analysis group, HDL was the lipoprotein cholesterol fraction that correlated best with MELD score, albumin level, bilirubin level, INR, and creatinine level. Kaplan–Meier plots are shown in Figure 2. In both the analysis and validation periods, patients with HDL levels less than 30 exhibited significantly poorer pretransplant survival (Figure 2A and D; each P ⬍ .001 by log-rank test). Patients with low HDL levels also were more likely to undergo transplantation within 1 year (Figure 2B and E). The likelihood of

289

death or transplantation at 1 year (Figure 2C and F) exceeded 60% in patients with HDL levels less than 30; by comparison, 80% of patients with HDL levels of 30 or more survived with cirrhosis for more than 1 year. By Cox proportional hazards analysis, in the analysis group, the hazard ratio for pretransplant death associated with HDL values less than 30 was 3.4 (95% confidence interval, 2.2–5.2, P ⬍ .01). In the validation group (Figure 2B) the comparable hazard ratio was 2.5 (95% confidence interval, 1.2– 5.2, P ⬍ .01). Hazard ratios for death or transplantation associated with HDL cholesterol levels less than 30 in the analysis and validation groups were, respectively, 3.7 (95% confidence level, 2.5–5.5, P ⬍ .001) and 3.4 (95% confidence level, 2.2–5.2, P ⬍ .001).

Discussion In this study, we showed that HDL cholesterol level, as measured inexpensively by routine automated laboratory methods, is a useful liver function test and an independent indicator of prognosis in patients with advanced cirrhosis. HDL cholesterol level correlated with other tests of liver function including serum albumin level, bilirubin level, and INR, as well as with a validated prognostic index, the MELD score. In multivariate analysis, low HDL cholesterol levels showed predictive value independent of MELD score for 6- and 12-month mortality in patients with noncholestatic cirrhosis. Total cholesterol level was also a good predictor, whereas VLDL cholesterol and LDL cholesterol levels were relatively poor indicators of liver function or prognosis. The prognostic value of low HDL cholesterol level was confirmed prospectively in patients referred after November 1, 2002. HDL levels less than 30 mg/dL were Table 4. Results of Multivariate Analysis (Binary Logistic Regression), Analysis Group 95% confidence interval for hazard ratio Interval 90 days 180 days 365 days

Independent predictors

Hazard ratio

Lower

Upper

P

MELD HDL MELD HDL

.839 1.062 .903 1.066

.772 1.012 .824 1.030

.912 1.115 .989 1.103

⬍.001 .014 .028 ⬍.001

NOTE. Variables analyzed included MELD score and total, LDL, HDL, and VLDL cholesterol levels. Outcome equals survival, censored at transplantation. The MELD score was calculated using the formula used by the United Network for Organ Sharing in assigning priority for liver transplantation (www.unos.org). The hazard ratio indicates the effect of a 1-unit decrease in the parameter of interest on odds for cirrhotic death. At 180 and 365 days, the HDL cholesterol level was found to be an independent predictor of mortality; no other lipoprotein fraction provided independent prognostic information. See text for details.

290

HABIB ET AL

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 3, No. 3

Table 5. Receiver Operating Characteristic Curve Analysis of Lipoprotein Fractions as Predictors of Mortality (Censored at Transplantation) in Analysis Cohort 95% confidence interval Time

Mortality

90 days

7.8%

180 days

10.7%

365 days

31.7%

Cholesterol C statistic fraction AUROC Total LDL HDL VLDL Total LDL HDL VLDL Total LDL HDL VLDL

.78 .65 .86 .76 .64 .49 .86 .69 .60 .46 .78 .62

P ⬍.001 .046 ⬍.001 ⬍.001 .039 .841 ⬍.001 .007 .085 .485 ⬍.001 .031

Lower Upper .66 .49 .79 .63 .50 .33 .75 .56 .49 .35 .69 .52

.91 .81 .94 .89 .79 .65 .96 .81 .71 .58 .87 .73

NOTE. Significance expressed relative to reference line (AUROC ⫽ .5). AUROC, area under receiver operating characteristic curve.

associated with a greater likelihood of transplantation within 1 year and a 3-fold increase in the hazard ratio for pretransplant death. It has long been recognized that total serum cholesterol level decreases in patients with cirrhosis, and that measurement of cholesterol may be of prognostic value. Milani et al2 found in 73 cirrhotic patients who died of liver failure that cholesterol level was associated inversely with survival. In patients hospitalized with spontaneous bacterial peritonitis, Llovet et al3 found that cholesterol was 1 of 7 parameters independently associated with short-term prognosis. D’Arienzo et al4 noted a progressive decrease in plasma cholesterol in 34 patients with viral cirrhosis, Child’s class C; all patients with total cholesterol level ⬍100 mg/dL died within 17 months, whereas 75% of those with cholesterol level ⬎100 mg/dL survived at least 2 years. A recent study by Zauner et al5 examined mortality in 196 cirrhotic patients requiring intensive care. They found that cholesterol level was associated independently with in-hospital mortality, along with bilirubin level, creatinine clearance, and lactate level. The levels of individual lipoproteins in patients with cirrhosis also have been studied, and these results in general support our finding that HDL cholesterol level is the most strongly associated with liver function. Skrede et al6 found a correlation between albumin level and ␣-lipoproteins in patients with chronic liver disease; Feher et al7 confirmed decreased ␣-lipoproteins and also noted decreased pre-␤ lipoproteins in patients with cirrhosis. Sabesin8 reported that HDL cholesterol level decreased in active alcoholinduced liver disease and returned toward normal with

abstinence. Rubies-Prat9 reported that HDL cholesterol and apolipoprotein A levels, measured by a precipitation method, were decreased in patients with cirrhosis, whereas HDL phospholipid concentrations were normal. In a study by Breier et al10 using rate zonal centrifugation and disc electrophoresis, patients with end-stage cirrhosis, compared with normal controls, exhibited marked decreases of total, HDL, and VLDL cholesterol, but not LDL cholesterol levels. In contrast, Cicognani et al11 reported that cirrhotic patients exhibited a decrease in LDL cholesterol as well as total and HDL cholesterol levels. The pathogenesis of the low HDL cholesterol level in patients with cirrhosis likely is multifactorial. Apolipoprotein A1, the major HDL lipoprotein, is synthesized and secreted by the liver, and synthesis of liver-specific proteins is impaired globally in cirrhosis. Poynard et al12 and Imbert-Bismut et al13 found that circulating levels of apolipoprotein A1 decrease with progression of liver fibrosis in alcoholism or hepatitis C, and decrease further with onset of cirrhosis. Malnutrition also may play a role; in patients with decompensated alcohol-induced cirrhosis and evidence of malnutrition, studied by Bories and Campillo,14 low levels of apolipoprotein A1 and HDL cholesterol were found to increase after a month of nutritional supplementation, accompanied by improvement in other parameters of liver function. A provocative study by Fabris et al15 suggested that decreased HDL levels in patients with hepatitis C preceded cirrhosis and correlated inversely with levels of interleukin 6. Thus, inflammatory mediators may influence hepatic secretion of nascent HDL.

Figure 2. Effect of low HDL cholesterol level on survival in patients with cirrhosis. Kaplan–Meier survival plots. (A–C) Analysis cohort (before November 1, 2002), (D–F) validation cohort (after November 1, 2002). (A and D) Pretransplant survival (patients censored at transplantation), (B and E) probability of transplantation as a function of time for surviving patients (hazard plot), and (C and F) survival until either death or transplantation. All differences were significant at P ⬍ .05 by log-rank test.

March 2005

Although our study excluded patients whose cirrhosis was caused by cholestatic disorders such as primary biliary cirrhosis or primary sclerosing cholangitis, available evidence suggests that similar decreases in HDL cholesterol levels accompany these conditions. Decreased apolipoprotein A1 and HDL cholesterol levels were noted in advanced primary biliary cirrhosis according to O’Kane et al.16 Tietge et al17 found that apolipoprotein A1 and HDL cholesterol levels decreased progressively with advancing stage of cirrhosis in both cholestatic and noncholestatic disease; in contrast, total cholesterol and apolipoprotein B levels were increased in patients with cholestatic liver disease. Nikkila18 reported that low levels of HDL cholesterol normalized after liver transplantation in patients with primary biliary cirrhosis. The accurate measurement of lipoproteins in cirrhosis may be problematic. Matas et al19 found that the Friedewald formula in cirrhosis consistently underestimated LDL cholesterol levels when compared with a gold standard of sequential preparative ultracentrifugation, reflecting an altered triglyceride/cholesterol ratio in VLDL in the cirrhotic patient. Thus, some of the lack of correlation of LDL and VLDL cholesterol with liver function and prognosis in cirrhotic patients may reflect inaccurate discrimination between the 2 lipoproteins. A systematic tendency to underestimate VLDL levels in patients with more advanced cirrhosis may have been responsible for the observed inverse correlation of VLDL cholesterol level with liver function in our study and could have masked, to some extent, a decrease in LDL cholesterol level. These considerations do not apply to measurement of either total or HDL cholesterol levels. Our study examined the association of lipoprotein cholesterol with prognosis, and has been validated prospectively. The independent association of HDL cholesterol level with 6- and 12-month mortality suggests that this parameter may be useful in identifying patients with deteriorating liver function who are at risk for terminal decompensation. Additional studies are needed to confirm this finding in other populations. Given the ready availability and reliability of this simple blood test, we propose that determination of HDL cholesterol level should be part of routine monitoring of patients with advanced noncholestatic cirrhosis.

References 1. Habib A, Abou-Assi SG, Mihas AA, et al. HDL cholesterol is an indicator of liver function and prognosis in advanced cirrhosis (abstr). Gastroenterology 2003;124:A-28. 2. Milani A, Marra L, Siciliano M, et al. Prognostic significance of

HDL IN CIRRHOSIS

3.

4.

5.

6. 7. 8. 9.

10.

11.

12. 13.

14.

15.

16.

17.

18.

19.

291

clinical and laboratory parameters in liver cirrhosis. A multivariate statistical approach. Hepatogastroenterology 1985;32:270 –272. Llovet JM, Planas R, Morillas R, et al. Short-term prognosis of cirrhotics with spontaneous bacterial peritonitis: multivariate study. Am J Gastroenterol 1993;88:388 –392. D’Arienzo A, Manguso F, Scaglione G, et al. Prognostic value of progressive decrease in serum cholesterol in predicting survival in Child-Pugh C viral cirrhosis. Scand J Gastroenterol 1998;33:1213–1218. Zauner C, Schneeweiss B, Schneider B, et al. Short-term prognosis in critically ill patients with liver cirrhosis: an evaluation of a new scoring system. Eur J Gastroenterol Hepatol 2000;12:517–522. Skrede S, Blomhoff JP, Elgjo K, et al. Serum proteins in diseases of the liver. Scand J Clin Lab Invest 1975;35:399 – 406. Feher J, Romics L, Jakab L, et al. Serum lipids and lipoproteins in chronic liver disease. Acta Med Acad Sci Hung 1976;33:217–223. Sabesin SM. Lipid and lipoprotein abnormalities in alcoholic liver disease. Circulation 1981;64:III– 84. Rubies-Prat J, Masdeu S, Nubiola AR, et al. High-density lipoprotein cholesterol and phospholipids, and apoprotein A in serum of patients with liver disease. Clin Chem 1982;28:525–527. Breier C, Lisch HJ, Braunsteiner H. Lipoproteins, HDL-apolipoproteins, activities of hepatic lipase and lecithin-cholesterol acyltransferase in the plasma of patients with post-alcoholic endstage liver cirrhosis. Klin Wochenschr 1983;61:929 –931. Cicognani C, Malavolti M, Morselli-Labate AM, et al. Serum lipid and lipoprotein patterns in patients with liver cirrhosis and chronic active hepatitis. Arch Intern Med 1997;157:792–796. Poynard T, Abella A, Pignon JP, et al. Apolipoprotein AI and alcoholic liver disease. Hepatology 1986;6:1391–1395. Imbert-Bismut F, Ratziu V, Pieroni L, et al. Biochemical markers of liver fibrosis in patients with hepatitis C virus infection: a prospective study. Lancet 2001;357:1069 –1075. Bories PN, Campillo B. One-month regular oral nutrition in alcoholic cirrhotic patients. Changes of nutritional status, hepatic function and serum lipid pattern. Br J Nutr 1994;72:937–946. Fabris C, Federico E, Soardo G, et al. Blood lipids of patients with chronic hepatitis: differences related to viral etiology. Clin Chim Acta 1997;261:159 –165. O’Kane MJ, Lynch PL, Callender ME, et al. Abnormalities of serum apo A1 containing lipoprotein particles in patients with primary biliary cirrhosis. Atherosclerosis 1997;131:203–210. Tietge UJ, Boker KH, Bahr MJ, et al. Lipid parameters predicting liver function in patients with cirrhosis and after liver transplantation. Hepatogastroenterology 1998;45:2255–2260. Nikkila K. Liver transplantation restores low serum levels of very low density and high density lipoproteins in end-stage primary biliary cirrhosis. Ann Med 1992;24:129 –136. Matas C, Cabre M, La Ville A, et al. Limitations of the Friedewald formula for estimating low-density lipoprotein cholesterol in alcoholics with liver disease. Clin Chem 1994;40:404 – 406.

Address requests for reprints to: Douglas M. Heuman, MD, GI Section (111-N), McGuire Department of Veterans Affairs Medical Center, 1201 Broad Rock Boulevard, Richmond, Virginia 23249-0001. e-mail: [email protected]; fax: (804) 675-5816. The authors thank Cecile Rock and Chauvonna Taylor for clerical assistance. The authors also thank Brenda Salvas, Liver Transplant Manager for the Department of Veterans Affairs, for her assistance in tracking outcomes of patients referred for consideration of liver transplantation.