Ultraviolet cutoff area and predictive ability of partial least squares regression method: A pharmaceutical case study.

UV cutoff area (COA) is known to be the wavelength band where solvents used for analysis can absorb radiation and accordingly affect the absorption spectra of drugs of interest being analyzed, even if blank experiments are done to eliminate solvent interference. However, this area may show peaks of significance for some drugs, and accordingly some researchers tend to include it in analysis. This study is presenting the importance of avoiding using COA, where it may represent significant negative effect on predictive ability of some linear chemometric methods like partial least squares regression PLSR. The presented study is using previously analyzed pharmaceutical mixtures of Dapoxetine Hydrochloride (DAP) and Tadalafil (TAD) as a case study, whether in pure forms or in dosage form, where the study uses two datasets for analysis, the first aims to include COA and the second dataset avoids it, then a statistical comparison is conducted for training sets, test sets and dosage form datasets to see how far COA may interfere with analysis results. Generally, the results show significant difference in datasets for t and F statistics for analysis of dosage form sets; which reflects changes in predictive ability of used chemometric method upon inclusion of COA in absorbance datasets, and accordingly unsuitability of using COA especially for routine quality control analysis of pharmaceutical mixtures.