02524nas a2200205   4500000000100000008004100001260001600042653002400058653002200082653001700104653002500121100001500146700001500161700002300176700001700199245009200216490000900308520198700317022001402304       2024                            d  bElsevier BV10aInorganic Chemistry10aOrganic Chemistry10aSpectroscopy10aAnalytical Chemistry1 aSantos JDS1 aDiniz PHGD1 aSoares Sobrinho JL1 aSoares MFDLR00aDapsone determination in tablets to leprosy treatment using a portable NIR spectrometer0 v12993 a<p>Leprosy is a neglected disease with limited technological advancements in the development of new formulations and analytical methods. The pharmaceutical industry offers several analytical alternatives for drug quality control; however, they are expensive, environmentally harmful, and labor-intensive. To address this, the United States Pharmacopoeia recommends the use of high-performance liquid chromatography (HPLC). In such a context, there arises a need for eco-friendly analytical approaches, such as near-infrared (NIR) spectroscopy, which offers the advantages of being non-destructive and eliminating the need for chemicals and solvents. Thus, in this study, we initially developed and validated an HPLC method for the quantification of&nbsp;dapsone&nbsp;in tablets. The method demonstrated selectivity, linearity, precision, accuracy, and very low limits of detection and quantification. Subsequently, the dapsone contents determined by HPLC in manipulated and industrial tablets were used as reference values to construct chemometric models based on&nbsp;NIR&nbsp;spectra acquired using a portable device. Partial Least Squares (PLS) and the Successive Projections Algorithm for interval selection coupled with PLS (iSPA-PLS) were then employed. Only the models constructed with linear baseline-corrected (LBC) spectra yielded significant results, with a ratio of performance to deviation (RPD) higher than 3. LBC/iSPA-PLS was the best model, selecting 3 out of 5 divided intervals, incorporated analytical information related to the aromatic rings and amino groups present in the dapsone molecule. This model achieved a relative prediction error of only 2.78 and 2.54&nbsp;% for both the manufactured and industrialized tablets, respectively, and a limit of detection of 5.38&nbsp;mg per 100&nbsp;g of tablet. Therefore, the proposed methodology can be hereafter applied to the quality control of dapsone tablets, even in the presence of non-modeled excipients.</p>
  a0022-2860