Relative Quantification Based on Logistic Models for Individual Polymerase Chain Reactions

The quantitative real-time reverse transcription polymerase chain reaction (RT-PCR) technology measures molecular variations in specific biomarkers. Relative quantification determines the target expression relative to an external standard or reference sample and should be adjusted for the PCR efficiencies actually achieved. More accurate methods of estimating PCR efficiency require a number of serial dilutions of the target sample, which is not generally feasible for clinical specimens. Alternatively, the efficiency of a single reaction may be estimated by considering kinetic data from this reaction. The current methods of estimating individual reaction efficiency require finding its exponential phase, which may affect the accuracy and precision of efficiency estimates. Thus, a model adequately representing all available kinetic RT-PCR data is preferable, but no such model is currently in use for relative quantification. In this work, we use a logistic model for all kinetic data from each RT-PCR and propose a new method of efficiency-adjusted relative quantification based on the estimates from the fitted logistic models. This method allows incorporating multiple replicates and possibly multiple reference ('housekeeping') genes for estimating relative expression and corresponding confidence interval. Real kinetic RT-PCR data are used to compare the proposed and standard methods. The methods are applied to the clinical data from the ongoing study of guanylyl cyclase C as a biomarker for colorectal cancer.