Chemometrics is the chemistry field responsible for planning and extracting the maximum of information of experiments from chemical data using mathematical tools (linear algebra, statistics, and so on). Active pharmaceutical ingredients (APIs) can form impurities when exposed to excipients or environmental variables such as light, high temperatures, acidic or basic conditions, humidity, and oxidative environment. By considering that these impurities can affect the safety and efficacy of the drug product, it is necessary to know how these impurities are yielded and to establish the pathway of their formation. In this context, forced degradation studies of pharmaceutical drugs have been used for the characterization of physicochemical stability of APIs. These studies are also essential in the validation of analytical methodologies, in order to prove the selectivity of methods for the API and its impurities and to create strategies to avoid the formation of degradation products. This review aims to demonstrate how forced degradation studies have been actually performed and the applications of chemometric tools in related studies. Some papers are going to be discussed to exemplify the chemometric applications in forced degradation studies.

The occurrence of polymorphic transitions is a serious problem for pharmaceutical companies, because it can affect the bioavailability of the final product. With several known polymorphic forms carbamazepine is one of the most problematic drugs in this respect. Raman spectroscopy is a vibrational technique that is becoming very important in the pharmaceutical field, mainly due to its highly specific molecular fingerprint capabilities and easy use as a process analytical tool. However, multivariate methods are necessary both for identification and quantification. In this work an analytical methodology using Raman spectroscopy and interval Partial Least Squares Regression (iPLS), was developed in order to quantify mixtures of carbamazepine polymorphs in the presence of the most common excipients. The three polymorphs CBZ I, CBZ III and CBZ DH (which is a dihydrate) were synthesized and characterized by PXRD and DSC. Subsequently, tablets were manufactured using excipients and 15 different mixtures of carbamazepine polymorphs. The iPLS model presented average prediction validation errors of 1.58%, 1.04% and 0.22% wt/wt, for CBZ I, CBZ III and CBZ DH, respectively, considering the whole mass of the tablet. The model presents a good prediction capacity and the proposed methodology could be used to perform quality control in final products.