Quantification of human serum insulin concentrations in clinical pharmacokinetic or bioequivalence studies: what defines the “best method”?

Background: Historically, quantitative clinical diagnostic assays (QCDAs) have not been accepted for use in pharmacokinetic or bioequivalence studies because they do not fully comply with the US Food and Drug Administration (FDA) Guidance for Industry: Bioanalytical Method Validation (e.g., full calibration curve not generated with each analytical run). Samples from a bioequivalence study were analysed for insulin and C-peptide concentrations with QCDAs and guidance-conforming radioimmunoassays (RIAs) and the results compared across and within assays. Methods: Serum samples (n=1913) from study MKC-TI-142 were analysed first using the Roche E170 electrochemiluminescence immunoassay (ECLIA) for insulin concentration and the Immulite 2000 chemiluminescence immunoassay (CLIA) for C-peptide, and then using the corresponding Millipore RIAs. Results: The insulin assays were highly correlated: r2=0.92 excluding samples requiring dilution and R2=0.88 including all samples. There was increasing negative bias of the ECLIA compared with the RIA with increasing insulin, especially with samples that required dilution for the RIA. The ECLIA had significantly fewer below-quantifiable-limit samples, a larger dynamic analysis range without dilution, and tighter agreement within incurred sample reanalysis (ISR) as compared with the RIA. The C-peptide assays showed good agreement but the CLIA method produced ISR results that were in closer agreement with the original values. Conclusions: The science indicates that the QCDAs are appropriate for the quantification of serum insulin (ECLIA) and C-peptide (CLIA) concentrations in human pharmacokinetic and bioequivalence studies even though the calibration curve is not generated in each analytical run