Comparison of UHPLC and HPLC methods for the assay of prostanoids: "are the methods equivalent in terms of accuracy and precision?"

Abstract

Background: As new methods are developed to increase efficiency and higher analytical performance, it is necessary to evaluate their quality in comparison to standard methods. To understand how the analytical performance changes between methods, it is common to compare the validation parameters; sensitivity, linearity, accuracy and precision. Here, we compare an UHPLC-UV method to the HPLC-UV method (reference method) for the simultaneous determination of seven prostanoids. Though the basic chromatography theory is the same for HPLC and UHPLC, the instrumentation has been modified to accommodate higher pressures, lower flow rates and smaller sample size. The differences in analytical instrumentation and procedures can give rise to method inequivalencies. Our approach evaluates the UHPLC and HPLC methods and poses the question: are the methods equivalent? To answer this question a statistical comparison of the analytical performance and method parameters is necessary.

Results: Statistical comparisons were performed using the t-test, F-test, regression analyses (ordinary linear regression and Deming regression) and Bland-Altman analyses. Statistical comparison of the results, suggested that the precision (amount of variability) is different (p < 0.05) for the HPLC and UHPLC methods. Whereas, the accuracy (method bias and the means) is similar (p > 0.05) for 8-isoprostane, 11-dehydro TXB₂, PGE₂ PGF(2α), PGD₂ and 15-deoxy Δ¹²,¹⁴ PGJ₂.

Discussion: Ordinary linear regression shows that the methods are well correlated for all compounds. The Deming regression, which assumes error in both the methods, suggests the existence of a proportional and constant bias for 11-dehydro TXB₂ and only proportional bias for 8-isoprostane, PGF(2α), PGD₂ and 15-deoxy Δ(12,14) PGJ₂ between the two methods. According to Deming regression, the two methods are statistically similar for 6-keto PGF(1α) and PGE₂. The Bland-Altman analyses indicate the two methods are commutable.