MELD fails to measure quality of life in liver transplant candidates

Authors


  • See Editorial on Page 134

Abstract

Previous studies have demonstrated an association between Child Turcotte-Pugh (CTP) class and impaired quality of life. However, the relationship between the model for end-stage liver disease (MELD) score and quality of life (QOL) has not been well studied. In this study, quality of life questionnaires (Medical Outcomes Short Form 36 [SF-36] and the Chronic Liver Disease Questionnaire [CLDQ]) were administered to 150 adult patients awaiting liver transplantation. We also collected demographic data and laboratory results and recorded manifestations of hepatic decompensation. The study found that all domains of the SF-36 and CLDQ were significantly lower in our patient cohort than in normal controls (P < .001). There was a moderate negative correlation between CPT class and physical components of the SF-36 (r = −.30), while there was a weak negative correlation (r = −.10) between CPT class and the mental component. There was a negative moderate correlation between CPT class and overall CLDQ (r = −.39, P < .001) and a weak correlation (r = −.20) between MELD score and overall CLDQ score. Both encephalopathy (correlation coefficient = −.713, P = .004) and ascites (correlation coefficient = −.68, P = .006) were predictive of the QOL using CLDQ (adjusted R2 = .1494 and f = 0.000). In conclusion, in liver transplant candidates, the severity of liver disease assessed by the MELD score was not predictive of QOL. The presence of ascites and/or encephalopathy was significantly associated with poor quality of life. CTP correlates better to QOL, probably because it contains ascites and encephalopathy. (Liver Transpl 2005;11:218–223.)

Liver allocation is currently based on the model for end-stage liver disease score (MELD).1 The transition from using Child-Turcotte-Pugh (CTP) class to the MELD score in liver transplantation was based on studies demonstrating that the MELD score better predicted short survival than did the United Network for Organ Sharing (UNOS) status, which was based on CTP score.2, 3, 4 In a recent study of 3437 adults added to the liver transplantation waiting list, the MELD score was found to be a better predictor of short-term survival than CTP scores.5

The MELD score uses three laboratory parameters—International Normalized Ratio (INR), serum creatinine, and serum bilirubin,2, 3—to predict short-term prognoses and thus is believed to be an accurate measure of liver disease severity. Complications such as hepatic encephalopathy and ascites, which are inherent in the CTP score, are not included in the MELD score.6 These manifestations of decompensated liver disease are subjective and lack strict operational definition, rendering them amenable to diverse interpretation. Moreover, these manifestations have not shown to increase the predictive ability of the MELD score substantially.4

While previous studies have demonstrated that liver disease severity as assessed by the CTP score is associated with impaired quality of life,7, 8, 9, 10, 11 it is not clear whether the MELD score is also associated with quality of life. Both hepatic encephalopathy and ascites, which can affect quality of life, are not part of the MELD score. Furthermore, the MELD score has not been correlated with the severity of ascites and hepatic encephalopathy.12 Thus, liver disease severity assessed by the MELD score may no longer correlate with quality of life. Thus, we studied the relationship between liver disease severity and quality of life (QOL). Our hypothesis was that the liver disease severity, as assessed by the MELD score, is associated with both liver-specific and general quality of life.

Abbreviations

MELD, model for end-stage liver disease; CTP, Child-Turcotte-Pugh; QOL, quality of life; SF-36, Medical Outcomes Short Form 36; CLDQ, Chronic Liver Disease Questionnaire; UNOS, United Network for Organ Sharing; INR, International Normalized Ratio; HRQL, health-related quality of life; PF, physical functioning; RP, role limitation–physical; BP, bodily pain; GH, general health; VT, vitality; SF, social functioning; RE, role limitation–emotional; MH, mental health; SF-36, Medical Outcomes Short Form 36; ABM, abdominal symptoms; FAM, fatigue; SYM, systemic symptoms; ACM, activity; EMM, emotional function; WOM, worry; CLDQ, Chronic Liver Disease Score.

Patients and Methods

Study Patients

Adult patients awaiting their first liver transplant at the University of California Los Angeles between July 2003 and May 2004 were enrolled. Patients were excluded if they were unable to provide consent, such as those with active hepatic encephalopathy. We collected demographic data including age, gender, marital status, employment status, and race. Liver disease etiology was also recorded. Laboratory data were obtained on the same day that questionnaires were administered. All recruitment and procedures followed Institution Review Board–approved protocol.

Disease Severity

Liver disease severity was assessed using the CTP classification and the MELD score.2, 3, 6 The MELD score was calculated by setting the minimum value of creatinine at 1 mg/dL and not including the liver disease etiology. We recorded the presence and treatment of ascites, variceal bleeding, encephalopathy, pruritus, and hepatocellular carcinoma.

Questionnaires

During an office visit, each patient completed two self-administered health-related quality of life questionnaires: the Medical Outcomes Short Form 36 (SF-36) and the Chronic Liver Disease Questionnaire (CLDQ). Both are validated questionnaires used to assess quality of life.

The SF-36 is a generic questionnaire that includes 36 items separated into 8 scales (physical functioning, role limitation–physical, bodily pain, general health, vitality, social functioning, role limitation–emotional, and mental health).17 Two summary scores can be obtained, a mental component and a physical component. Range for total scores is 0 to 100; for physical component, 8 to 73; and for mental component, 10 to 74. A higher score indicates a higher QOL, while a lower score indicates a lower QOL. Raw scores were transformed to a scale of 0 to 100.17 Scores for the general United States population were obtained from previously published data.7, 18

The CLDQ is a quality of life questionnaire specific for liver disease that was developed using focus group input, expert opinion, and factor analysis.9 The CLDQ includes 29 items separated into 6 compartments (abdominal symptoms, activity, emotional function, fatigue, systemic symptoms, and worry). Scores for CLDQ range 1 to 7.19 As with the SF-36, both summary scores and domain scores can be found, and a higher score indicates a higher QOL.

Statistics

Continuous variables were presented as means (± standard deviation [SD]). We used a Pearson correlation coefficient for continuous variables and a Spearman correlation coefficient for categorical variables.20, 21 Means were compared using t-tests for independent groups. Regression analysis was used to evaluate predictors of higher QOL score. Predictors included gender, ethnicity, marital status, employment status, etiology of liver disease, encephalopathy, variceal bleeding, ascites, and MELD score. We tested multiple regression models using each domain of the CLDQ and SF-36 as an outcome variable with the same list of predictor variables. Standard model checking techniques were used. A P value < .05 in a 2-tailed test was considered statistically significant. Statistical analyses were performed using STATA (STATA Corporation, Tex, US).

Results

Patients

The demographic characteristics of the study population are shown in Table 1. Most patients were male, non-Hispanic, white, and had cirrhosis from hepatitis C. The mean (± SD) MELD score was 13.56 (± 6.28). Thirty patients were classified as CTP class A, 71 as B, and 49 as C. Thirteen percent of the patient cohort had hepatocellular carcinoma. Manifestations of decompensation were found in 73% and included ascites (65%), hepatic encephalopathy (41%), and bleeding gastroesophageal varices (29%). Of the patients with ascites, 72 (48%) were on diuretics and 17 (11.3%) were considered refractory to medical treatment. Eight patients had a history of spontaneous bacterial peritonitis. A prior TIPS procedure had been performed in 14 patients; none had undergone a portasystemic shunt. Fifty (33%) patients were on lactulose, 7 (.5%) were on neomycin, and 16 (11%) were on both lactulose and neomycin. Seven (0.5%) patients required quinine treatment for cramps. Patients with hepatitis C were not on interferon therapy. There was a moderate to strong correlation between CTP class and MELD score (r = .68, P < .001).

Table 1. Clinical Characteristics of Study Population
Age (year)52.9 ± 9.23
Men/Women94/56
Ethnicity 
 Non-Hispanic White65
 Hispanic White60
 African American8
 Asian17
Employment (yes/no)51/99
Etiology of Disease 
 Hepatitis C77 (51%)
 Hepatitis B19 (13%)
 Alcohol18 (12%)
 Nonalcoholic steatohepatitis11 (7%)
 Primary Biliary Cirrhosis8 (5%)
 Cryptogenic6 (4%)
 Primary Sclerosing Cholangitis4 (3%)
 Autoimmune Hepatitis3 (2%)
 Hemochromatosis2 (1%)
 Sarcoidosis1 (.7%)
 Drug-induced1 (.7%)

Quality of Life Domains

All domains of the SF-36 scores were significantly lower in our patient cohort than in normal controls (P = .0001) (Figure 1).17 There was a negative moderate correlation between overall SF-36 scores and CTP class (r = −.40). The greatest differences were found in the physical health domains (physical functioning, role limitation–physical, and bodily pain). There was a moderate negative correlation between CTP class and the physical components of the SF-36 (r = −.30). Moderate correlations were found between physical functioning (r = −.44), role limitation–physical (r = −.46), bodily pain (r = .30), and general health (r = .20) domains and the CTP class. There was a negative weak correlation (r = −.10) between CTP class and the mental component of the SF-36. There was almost no correlation between CTP class and vitality/social functioning (r = .09); there was moderate correlation between CTP class and role limitation–emotional (r = −.38) and mental health (r = −.43).

Figure 1.

Mean values of Medical Outcomes Short Form 36 domains in patients with cirrhosis and United States normal controls.17P value was less than .001 for all domain comparisons between patients with cirrhosis and United States normal controls.

There was weak to moderate correlation between SF-36 score and MELD score (r = −.26). Between MELD score and the physical components of the SF-36, the correlation was weak (r = −.19). A weak correlation (r = −.20) was also found between MELD score and physical functioning and role limitation–physical. Almost no correlation was found between MELD and bodily pain and general health (r = −.07). There was almost no correlation (r = −.03) between MELD score and the mental component of the SF-36. There was almost no correlation between MELD scores and vitality (r = −.013), social functioning (r = −.012), and role limitation–emotional (r = −.094). There was a weak correlation between MELD score and mental health (r = .17).

All domains of the CLDQ were significantly lower in our patient cohort than in normal controls (Figure 2).7 There was a negative moderate correlation between CTP class and overall CLDQ (r = −.39, P < .001). We found moderate correlation between CTP class and several domains in the CLDQ such as abdominal symptoms (r = −.38), fatigue (r = −.43), systemic symptoms (r = −.31), activity (r = −.35), and emotional function (r = −.37). There was a weak to moderate correlation between CTP class and worry domain (r = −.27). We found a weak correlation between MELD scores and abdominal symptoms (r = −.22), fatigue (r = −.21), systemic symptoms (r = −.31), activity (r = −.24), emotional function (r = −.18), and worry (r = −.11).

Figure 2.

Mean values of Chronic Liver Disease Score domains in patients with cirrhosis and United States normal controls.7P value was less than .001 for all domain comparisons between patients with cirrhosis and United States normal control.

Predictive Variables

Univariate analysis indicated that MELD score was not a significant predictor of CLDQ (P = .97) or SF-36 (P = .96). In the regression analysis mode, both encephalopathy (regression coefficient = −.713, P = .004) and ascites (correlation coefficient = −.68, P = .006) were predictive of quality of life using CLDQ (adjusted R2 = .1494 and f = 0.000). However, no independent predictors could be found for QOL using SF-36 (overall model f = .97). Gender, ethnicity, marital status, employment status, etiology of liver disease, encephalopathy, variceal bleeding, ascites, and MELD score were not predictive of SF-36 score. We found that neither MELD score nor CTP class was an independent predictor of any of the domains of CLDQ and SF-36 when tested separately (P > .05).

Discussion

Patients consider quality of life in addition to survival when making healthcare decisions. Since the early 1970s, there has been an increased interest in measuring the impact of many modern interventions, not only in terms of survival but also in terms of quality of life. In liver disease in particular, it is essential to understand the impact of the disease on quality of life because of the prolonged wait for transplantation. Organ allocation for liver transplantation is currently entirely based on the MELD score, an objective measure of liver disease severity without input from QOL measures. Our results confirm previous observations that quality of life is impaired in patients with end-stage liver disease. However, we found a poor correlation between both generic and disease-specific quality of life scores and disease severity measured by MELD scores.

Many instruments and techniques have been developed to assess quality of life, and several of them were designed for use in patients with chronic liver diseases. The CLDQ has been validated in all chronic liver disease types.9, 22, 23 On the other hand, the 36-item Short Form Health Survey (SF-36) is a generic instrument used primarily in general populations and has an advantage of a comparison across different disease populations.23, 24 Our results suggest that a moderate correlation exists between CTP class and SF-36 scores. In particular, CTP class was correlated with the physical components of the SF-36 (physical functioning, role limitation–physical, and bodily pain). On the other hand, the MELD score was only weakly correlated with the physical components. Similarly, there was moderate correlation between CTP class and several domains of the CLDQ but only a weak correlation was found between the MELD score and the overall CLDQ score.

One of the purposes of the MELD for organ allocation is to objectify the criteria defining disease severity. Unlike the CTP, the MELD score excludes ascites and encephalopathy. The MELD score itself does not correlate well with severity of hepatic encephalopathy or ascites.12 In our model, encephalopathy and ascites were found to be important factors to consider in measuring QOL. CLDQ scores decreased by a factor of .7 when encephalopathy existed and by a factor of .14 when ascites were present. This fact is not surprising considering that both of these manifestations likely interfere significantly with day-to-day functions of affected individuals.

Our study has several important limitations. First, the results are not likely generalizable to all patients with cirrhosis. Patients in our study were listed for transplantation after screening by psychiatrists, social workers, hepatologists, and surgeons. Patients with depression are typically listed for liver transplantation only after complying with recommended pharmacologic treatment and/or psychotherapy. Moreover, all patients in our study were enrolled in outpatient clinics. Patients with more severe liver disease may not have been captured. Nevertheless, our patients reflect those seen in outpatient pretransplant offices and clinics. Another limitation is that patients with severe encephalopathy could not be enrolled in the study because of the inability to give consent and answer the questionnaires. Indeed, encephalopathy is a major predictor of QOL. But because encephalopathy is not used for MELD scoring, this limitation is unlikely to alter the association between MELD scores and quality of life. An additional limitation is the use of an historic control. The control group was obtained from previous studies.7, 17 However, we believed it was important to demonstrate the differences in quality of life between patients with liver disease and the typical general population. Future studies on quality of life in patients with liver disease should attempt contemporary controls.

Our study also has several important limitations involving the QOL instrument and our study population. The QOL instrument, originally validated against the CTP score, has not been validated against the MELD score or in different ethnic groups. Moreover, our study design was a cross-sectional survey, and the CLDQ has not been validated at different time points against objective variables. The correlation between QOL scores and worsening of liver disease requires further study. There are also limitations in our study population. Most of the study patients had liver disease from hepatitis C, thus limiting the generalization of our results to other disease states. Although patients with severe depression not responsive to standard antidepressants were likely excluded from transplant consideration, it is possible that patients may have developed depression after their evaluation. Equally important, depression was not formally assessed using criteria as from the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV-TR).25 Future studies on QOL involving patients with cirrhosis should include a formal assessment of depression.

CTP class previously has been shown to correlate with health-related quality of life (HRQL).7, 18 Younossi et al. used both CLDQ and SF-36 questionnaires to assess HRQL in 353 patients with chronic liver disease, failing to demonstrate the difference in the effect of various causes of liver disease on HRQL. However, the study did find an association of HRQL impairment with disease severity as measured by presence of cirrhosis.7, 8, 18 In our study, we found a moderate negative correlation between CLDQ and CTP scores (r = −.39). Our patient populations differed from those in previous reports because all our subjects met minimal listing criteria for transplantation and did not include patients with chronic liver disease without cirrhosis. It has been reported that patients with very advanced cirrhosis (CTP classes B and C) have profound impairment in their QOL. In these patients, detection of a difference in QOL between patients with CTP class B and CTP class C may not be possible.7, 8, 18, 26 This so-called “floor effect” in QOL scores was observed in our patients who were mostly CTP class B or C (80%). This data is consistent with previously reported QOL data in patients with chronic liver disease.8, 26

The results of our study suggest that QOL is not associated with liver disease severity as measured by the MELD score. Thus, liver disease severity should not be assumed to be a surrogate marker of QOL. Future studies in patients awaiting liver transplantation should include QOL measures as well as survival. Although patients may have low MELD scores, their quality of life may be impacted by liver disease, particularly if ascites and encephalopathy are present. Assessing the impact of liver disease severity on quality of life may need to be studied using additional validated instruments.

Acknowledgements

The authors thank Joshua Marehbian and Lisette Peralta for editing and David Ly, Nami Chun, and Yaeng Xayavong for administrative assistance.

Ancillary