Challenges in the measurement of plasma creatinine on the Roche cobas c702

The aim of this study was to document the differences between two Roche creatinine measurement methods, the CREP2 test on the cobas c702 and the CREA PLUS test on the Modular P. Samples with creatinine isotope dilution mass spectrometry (IDMS) values were analyzed on both instruments. Method comparison using the remaining plasma samples was performed twice, using two different lot numbers of reagents and two different lot numbers of calibrators on both instruments. Medians and percentiles of the plasma creatinine values produced on the Modular P and cobas c702 from 2012 to 2017 were compared. The recovery of samples with IDMS creatinine values (SRM 967a level 1, SRM 967a level 2, serum X and five serum pools from Roche) was 101.7%-110.2% on the cobas c702 and 98.9%-102.6% on the Modular P. Comparison of the two methods showed that the slope was close to 1.0 using linear, Deming and Passing Bablok regressions, but all equations showed a negative intercept, indicating that the cobas c702 overestimates plasma creatinine in relation to the Modular P by 4-6 µmol/L. The median value for routine plasma creatinine lies between 74 and 77 µmol/L for the Modular P and 81 µmol/L for the cobas c702. After the cobas c702 was factorized in September 2016, the median plasma creatinine value decreased to 75 µmol/L. In conclusion, the CREP2 method on the cobas c702 overestimates creatinine by 4-6 μmol/L, which has a significant influence on the estimated glomerular filtration rate (eGFR) in children.