Multivariate analysis to separate the signal given by cross-reactants in immunoassay with sample matrix dilution.

This paper describes a new approach to achieve selectivity in an immunoassay by separating the signals given by two cross-reactive compounds present simultaneously in a complex sample matrix. The method is based on the sequential dilution of the sample containing a mixture of the two analytes, spiking each diluted sample with a reference compound, and the detection by enzyme-linked immunosorbent assay (ELISA). The obtained multivariate response was used for the individual calibrations of the assay for each of the two cross-reactants simultaneously by using principal component analysis (PCA) and partial least squares regression (PLSR) data modeling. The calibration models showed that the signal separation due the analytes 2,4-dinitrophenol (2,4-DNP) and 4-nitrophenol (4-NP) was possible with a prediction concentration error of 1.4 microM and 72 microM, respectively.