Model-based strategy for bioanalytical method comparison: measurement of a soluble ligand as a biomarker

Abstract

Ligand binding assays (LBAs) are often the method of choice for quantification of protein biomarkers and therapeutic biologics during drug development. Soluble ligand X is a glycoprotein. To understand the role of circulating ligand X in drug-target relationship, an analytical method (Method 1) was developed and validated to measure circulating ligand X and to support early clinical studies. Change in the detection reagent led to the development and validation of a second method (Method 2). Both methods measure total circulating ligand X levels. To ensure that the method specificities and data were consistent upon method change, the two methods were cross-validated using three distinct sample types: (1) recombinant ligand X (rLIGX) spiked in buffer, (2) authentic serum samples containing endogenous ligand X (eLIGX), and (3) serum samples collected from patients being dosed with the therapeutic antibody (incurred samples). Methods were considered comparable if the 90% confidence interval (90% CI) fell within 0.80-1.25 for all sample types. The results from the comparison reveal that two methods were comparable for rLIGX samples with the 90% CI of 0.90-1.07. However, with eLIGX samples, Method 1 produced higher mean (± SD) concentrations 675 (± 316 pg/mL) than Method 2 195 (± 97 pg/mL) and the two methods were considered not comparable as the 90% CI was 0.27-0.29. With the incurred samples, the comparison results also indicated the incomparability of these two methods as the 90% CI was 0.57-0.65. To describe the statistically relevant relationship between two methods in analyzing the serum samples, linear and quadratic regression models were applied to derive two conversion equations; one each for eLIGX and incurred samples. The applicability of the equations was verified with independent study data to indicate that the equations can be used to relate two different sets of study data. A model-based strategy presented here can serve as an explicatory paradigm for other analogous situations in the future.