Non-linearity and gaps in results distribution over successive Roche Lipase method applications: improvement but persistence.

Non-linearity within the primary measurement range of a lipase assay (<300 U/L) has been shown on Cobas® Roche analyzers, causing gaps in results distribution between 300 and 400 U/L. Since, new lipase method applications (LMAs) have been used. The purpose is to retrospectively evaluate their impact on relative frequencies of lipase results (RFLs).Plasma lipase results from two hospital laboratories, assayed over 7.2 years, were collected. Over this period, three successive LMAs, characterized by automated repeat-on-dilution (1/11, 1/2, or 1/10), were applied for lipase results >300 U/L: LMA1 and LMA2 on the Modular®P800, Cobas®c501 and Cobas®C701 analyzers, and LMA3 on the Cobas®C701. RFLs were determined, linearity tests were performed, and inter-agreements between lipase results corrected and uncorrected for nonlinear biases were assessed, using 180 U/L as a decisional cut-off for acute pancreatitis.Overall, RFL gaps narrowed from LMA1 (300 to ∼380 U/L) to LMA3 (300 to ∼330 U/L). For a lipase activity fixed at 300 U/L, non-linearity biases were determined at -11.2% on the Modular®P800 (LMA1), -20.8% on the Cobas®c501 (LMA1), and -3.5% (LMA2) and -2.2% (LM3) on the Cobas®C701. Diagnostically, a maximum of 0.48% lipase results were misclassified as negative (LMA1 on the Cobas®c501), and a minimum of 0.01% misclassified as negative (LMA3 on the Cobas®C701). In conclusion, successive Roche lipase method applications improved linearity within the primary measurement range. While persisting, gaps in lipase results distribution narrowed with the evolution of the methods, with a minor impact in terms of diagnostic of acute pancreatitis.