Clinical use of plasma lactate concentration and evaluation of the analytical performance of two benchtop analyzers for the measurement of L-lactate in canine plasma

Abstract

Lactate has been used in small animal veterinary medicine since the 1990’s. Ongoing reviews of the literature had revealed that an up to date, comprehensive review of the use of lactate in small animal clinical practice was overdue. Consequently, this became the first objective of this thesis (Chapters 1 and 2). Through this review process, a surprising gap was identified in the veterinary literature. Despite widespread integration into clinical practice and validation of hand-held lactate analyzers against benchtop methods, there was no actual published evidence validating these benchtop methods for use in dogs (Chapter 3). Accordingly, this became the objective of the experimental component of this thesis (Chapter 4).

A partial method validation study was designed to satisfy requirements outlined by the American Society of Veterinary Clinical Pathologists (ASVCP). The objective was to evaluate the analytical performance and equivalency of two commercially available benchtop lactate analyzers: a high-end point-of-care analyzer, the Radiometer ABL 800 (ABL800), and a clinical laboratory analyzer, the Cobas Integra 400 (Cobas400). As a gold-standard, definitive method for lactate measurement is poorly defined, we elected to use canine plasma manually spiked with reference standard sodium lactate as our test medium. After a preliminary assessment of linearity, intra-assay precision, and inter-assay precision using manufacturer controls, we used the spiked plasma to assess linearity, intra-assay precision, accuracy, range and equivalency of the two analyzers.

Both analyzers demonstrated excellent precision. The Cobas400 was accurate over a wider range than the ABL800, however the ABL800 was more accurate within the most clinically relevant range. Method comparison was performed using the Cobas400 as the nominated reference method. Passing-Bablok linear regression and Bland-Altman analysis revealed that although the methods were in agreement, with ≈ 95% of measured differences falling within ± 1.96 standard deviations of the mean difference, they cannot be considered equivalent due to the presence of a small but clinically significant amount of constant and proportional bias.

In conclusion, both analyzers demonstrated adequate analytical performance for use in clinical practice and future validation studies. However, they cannot be considered equivalent due to the presence of both constant and proportional bias and should not be used interchangeably. Consequently, analyzer specific reference intervals should be applied, and the same analyzer should be used when serially trending lactate concentrations in an individual patient.