Background/aims: The model of end stage liver disease (MELD) includes serum creatinine, which is a poor surrogate marker of renal function in patients with cirrhosis. Especially in women and patients with advanced disease creatinine underestimates true renal function. Our objective was to assess whether or not the substitution of creatinine by cystatin C improves the prognostic performance of the model.
Methods: The association between MELD parameters and cystatin C with survival was investigated using a Cox proportional hazards model. A cystatin C-based MELD score was calculated from the results and compared with creatinine-based MELD in terms of discrimination and calibration.
Results: Four hundred and twenty-nine patients were included in the study; 19% died and 12% underwent liver transplantation during a median follow-up of 602 days. In multivariate Cox regression, cystatin C was an independent predictor of 90-day mortality with a hazard ratio of 8.0 (95% CI: 2.2-29.6). The median cystatin C-based MELD was 15, the median creatinine-based MELD was 12. Calibration and discrimination for 3 month and 1 year mortality was similar between the scores (AUC > 0.85 for both scores). Gender differences in cystatin C-based MELD were less pronounced than those in the creatinine-based model, because creatinine but not cystatin C was affected by gender.
Conclusion: Substitution of creatinine by cystatin C does not improve the predictive power of MELD.
© 2012 John Wiley & Sons A/S.