Evaluation of unfolded-partial least-squares coupled to residual trilinearization for four-way calibration of folic acid and methotrexate in human serum samples

The combination of unfolded-partial least-squares (U-PLS) with a recently proposed separate procedure, known as residual trilinearization (RTL), has been successfully employed for four-way data calibration. The chemometric method employs the evolution of excitation–emission matrices (EEMs) with time, for the resolution of folic acid–methotrexate mixtures, in human serum samples. The fluorogenic products monitored correspond to the oxidation of the studied analytes with potassium permanganate, in slightly acidic medium. The reaction is developed in 7 min and followed using a fast-scanning spectrofluorimeter, capable of recording each complete EEM in 12 s. This allows the acquisition of 10 successive EEMs, at different reaction times, during the development of the oxidation reaction, given rise to the four-way data set employed. The procedure, which had been previously reported for urine determination, is extended to serum analysis in this work. The combination of U-PLS/RTL is providing enhanced predictive results in comparison with standard methods as PARAFAC and N-PLS, in the presence of human serum, where significant unexpected components and or inner filter effects may occur.