Predicting outcome in patients with cirrhosis following acute decompensation: Can we do better?

Authors


Associate Professor Ed Gane, New Zealand Liver Transplant Unit, PO Box 92024, Auckland, New Zealand. Email: edgane@adhb.govt.nz

An estimated one billion people worldwide have cirrhosis, the majority of whom live within the Asia–Pacific region. Although most patients with compensated cirrhosis remain well, the development of complications is associated with a dramatic reduction in survival. One and 2-year mortality increases from 5% and 10% in compensated cirrhotics to 20% and 40% in those with decompensated cirrhosis.1 Some complications such as spontaneous bacterial peritonitis and hepatorenal failure are associated with less than a 50% 1-year survival. The only accepted therapeutic intervention in patients with decompensated cirrhosis is liver transplantation but this is not widely available or affordable in many countries in the Asia–Pacific region.

Over the past 10 years, several large retrospective series have evaluated baseline predictors for survival in patients with cirrhosis. It was hoped that identification of reliable predictors would enable the formulation of minimal listing criteria, which, if adopted by organ allocation systems, would prevent unnecessary transplantation (if carried out too early), or futile transplant (if carried out too late), thereby reducing waiting list mortality and facilitating equal access to organs across all transplant centers. The initial prognostic model for chronic liver disease was the Child–Turcotte–Pugh (CTP) score, incorporating serum bilirubin and albumin, prothrombin time, clinical ascites and encephalopathy. This is a crude predictor of survival in patients with cirrhosis.2,3 CTP was adopted by the United Network for Organ Sharing (UNOS) in 1998 to provide a transparent and fair organ allocation system for elective liver transplantation. Shortcomings of the CTP score are that it includes a subjective assessment of ascites and encephalopathy scores, prone to interobserver variability.

In contrast to the Child–Pugh score, the Model for End-Stage Liver Disease (MELD) is a purely objective prognostic model, incorporating three simple objective measurements: creatinine, bilirubin and the international normalized ratio (INR; (see).4 When applied to patients listed for liver transplantation, the MELD score accurately predicted 3-month waiting list mortality.5 In February 2002, UNOS replaced the CTP score with the MELD score for waiting list prioritization. However in a recent systematic review, only four of 11 studies comparing MELD to Child–Pugh score reported that MELD had superior predictive value for short-term waiting list mortality.6 Despite this, the institution of MELD has successfully reduced waiting list registrations (as MELD scores less than 15 are unlikely to receive a graft) and also waiting list mortality (as sickest patients are now prioritized). The introduction of MELD also appeared to be beneficial in single centers not participating in organ sharing networks.7

MELD has been modified in an effort to improve prognostic accuracy further. Change in MELD over a 30-day period (Δ MELD) is a more accurate predictor of waiting list mortality than a single measurement.8 The addition of serum sodium to the MELD model also increases prognostic accuracy, especially in those with lower MELD scores (< 21).9–11 Although MELD is best known for its role in organ allocation, it may also predict mortality in patients with early cirrhosis, prior to listing for transplantation.12 MELD score was initially derived to predict post-procedure mortality following transjugular intrahepatic portosystemic shunt (TIPS).4 MELD also predicts short-term morbidity and mortality in cirrhotic patients following cardiac and other non-hepatic surgery.13,14

Of note, the increased mortality following major surgery is largely due to increased deaths from sepsis and multiorgan failure rather than hepatic decompensation. The negative impact of extrahepatic organ failure would also explain why non-liver-specific physiological scoring systems such as Acute Physiology and Chronic Health Evaluation (APACHE) II and III are superior to MELD and CPS as predictors of short-term outcome in cirrhotic patients admitted to intensive care units following acute decompensation.15

In patients with acute liver failure, baseline predictors can identify those unlikely to survive on medical management alone and thereby facilitate early listing for emergency transplantation. Of note, the prognostic accuracy of the most widely used Kings' College Criteria have recently been improved by the addition of factors reflecting extrahepatic markers of extrahepatic organ failure, namely serum phosphate and lactate.16,17

In the paper by Cholangitas et al. in this issue of the Journal of Gastroenterology and Hepatology,18 the prognostic accuracy of two liver-specific prognostic models (MELD and CTP) are compared to three prognostic models derived from general intensive care patients (APACHE II, Sequential Organ Failure Assessment [SOFA] and Failing Organ Systems [FOS]) in cirrhotic patients admitted to the intensive care unit (ICU) following acute decompensation. The liver-specific models incorporate measurements of liver synthetic function which indirectly reflect the body's resistance to infection (complement, opsonizing proteins, chemotaxins etc.), whereas the general-ICU-specific prognostic models measure physiological disturbances seen in critical illness; namely renal failure, respiratory failure, hypotension and acidosis. Therefore, whereas MELD and CTP may be more useful in predicting mortality over a 3- or 12-month period, the general ICU scores reflect immediate mortality. These observations confirm those of Zauner et al.15

It is not surprising that the prognostic accuracy of all scores is improved by a delay in measurements until 48 h after admission to the ICU, given the improvement of many physiological parameters (acidosis, renal dysfunction, hypotension) following correction of hypovolemia.

The results from this current study demonstrate that advanced extrahepatic organ dysfunction significantly impacts short-term survival in cirrhotic patients with acute decompensation. In these patients, multiorgan failure is associated with universally poor outcome despite optimal medical management. However, emergency liver transplantation in this setting is not justified, with 1-year post-transplant survival rates of only 25%.19

In conclusion, following acute decompensation in patients with cirrhosis, the outcome is determined by the severity of extrahepatic organ dysfunction. General physiological scoring systems should be used to determine prognosis rather than the liver-specific MELD and CTP scores. However, patients with established multiorgan failure, despite 48 hours of intensive care, are unlikely to survive with either medical therapy or transplantation. Therefore, monitoring with general physiological scoring systems in cirrhotic patients following acute decompensation should be used to help guide management decisions and avoid futile escalation of therapy.