Application of Different Interval Variable Selectors for Quantification of Spectrally Overlapping Pharmaceuticals by Multivariate Calibration

The novel application of interval variable iterative space shrinkage approach iVISSA and partial least squares PLS calibration for quantification of three overlapping pharmaceuticals (paracetamol, guaifenesin, and phenylephrine) is presented in this work. In addition to spectral overlapping, the drugs are available in the commercial tablet in varying proportions where paracetamol and guaifenesin are 20 to 50 times higher than phenylephrine. Net analyte signal calculations indicated that pH has an high influence on drug overlapping, and the optimum pH was at 12.0, with a total overlapping 57-77% among solutes. To eliminate the influence of excipients on PLS calibration, all standard mixtures were prepared with a 5% level of excipients. For Guaifenesin, variable selection by NAS resulted in better PLS prediction. iVISSA, which developed under the condition that locations, widths, and combinations of selected intervals are optimized simultaneously, resulted in the excellent prediction of phenylephrine by selecting 34 spectral data in which the solute became intensely absorbed. Analysis of commercial tablet (250 mg paracetamol, 100 mg guaifenesin, 5 mg phenylephrine) with the help of NAS revealed that GUA was predicted with high accuracy (98.4%) and precision (RSD 3.9) using the range 225-239 nm. For phenylephrine, the selected intervals were 200-207, 224-225, 232-246, 250-254, 263, 267-270, 275, and 278-280 nm by iVISSA, which resulted in accurate quantification with high accuracy (104.3%) and precision (RSD 0.5). For paracetamol, including the entire range (200-300 nm, 101 points) was necessary for better PLS prediction while variables selection by NAS or iVISSA negatively affected PLS calibration for that drug. The accuracy and precision of the proposed method were validated against liquid chromatography and both methods were found statically comparable.