Background The Model for End Stage Liver Disease (MELD) score is utilized as a prognostic tool
for end-stage liver disease. The variations in laboratory tests have a substantial
impact on MELD. For creatinine as one parameter in the MELD score, measurements are
mostly performed by either Jaffé kinetic or enzymatic assays. In both assays, bilirubin
is recognized to significantly interfere with the creatinine detection.
Aim Our objective was to characterize the bias of bilirubin interference on creatinine
for both methods employing Roche cobas6000 clinical chemistry analyzers. Secondarily,
we aimed to investigate the impact of this interference on MELD.
Methods We set up two dilution matrices using water and PBS as solvents containing creatinine
ranging from 0 to 5 mg/dL and bilirubin ranging from 0 to 35 mg/dL. The effect of
bilirubin on creatinine levels was modeled by three-dimensional regression analysis
for both methods separately. Subsequently, the models were used to recalculate MELD
scores from retrospective laboratory results (N=13,186) from patients with liver cirrhosis.
Results With increasing bilirubin, creatinine by Jaffé was increased, while creatinine by
enzymatic method was decreased. Recalculation of MELD scores using corrected creatinine
values yielded a significant number of MELD scores which were up to two points lower
for Jaffé (N=564, 3.5%) and up to three points higher for enzymatic assay (N=1,016,
7.7%).
Conclusion The method-dependent interference of bilirubin on creatinine significantly impacts
MELD. For the calculation of standardised MELD scores, correction of creatinine values
accounting for both methods and bilirubin effect is required.
