Issue link: https://www.e-digitaleditions.com/i/1331851
eye on excipients 38 January/February 2021 Tablets & Capsules This edition of Eye on Excipients dis- cusses how multivariate analysis can be used to de-risk the use of excipients. The article describes a study that uses principal component analysis (PCA) to evaluate the use of anhydrous lactose manufactured from a second production line. The study compares batches repre- senting the historical variation for both production lines. The resulting data set can be useful for pharmaceutical com- panies as part of their control strategy when qualifying anhydrous lactose from the second production line. The analyses also show the strengths and challenges associated with PCA. Very small changes and patterns are highlighted, even when all data are well within specification limits. It is the authors' strong recommendation to have excipient suppliers regularly perform PCA to avoid trends in the data or in the excipients' quality attributes that are not understood. Pharmaceutical excipients are sub- stances other than the active pharma- ceutical ingredient (API) or prodrug that are included in the manufactur- ing process or contained in a fin- ished pharmaceutical dosage form [1]. Functional excipients enable an API to be processed into a final dos- age form and may enhance stability, bioavailability, or patient acceptance [2]. Excipient quality and consis- tency is important to ensure the con- sistent performance of every final dosage form. All production processes, includ- ing excipient production, have some inevitable degree of variation [3]. Variation in a process does not mean that the process is out of control, only that not all batches are exactly the same. The impact of this varia- tion on the final dosage form should be understood. Many authors have recognized the impact of batch-to- batch and vendor-to-vendor excip- ient variation and inconsistencies. Haware, et al. (2010) showed that physical properties between micro- crystalline cellulose (MCC) 101 grades and their functionality can be different for different vendors [4]. Gamble et al. (2010) reported batch- to-batch and vendor-to-vendor vari- ation in anhydrous lactose and sub- sequent impact on its processability and functionality [5]. In recent years, there has been a continuous drive from pharmaceu- tical companies and regulatory bod- ies to develop more robust pharma- ceutical formulations and processes based on knowledge [6]. This has resulted in requests to excipient suppliers to help de-risk the use of their excipients in line with Quality by Design (QbD) principles. Tra- ditionally, product quality in man- ufacturing is handled reactively by restricting flexibility in the manu- facturing process and by end-prod- uct testing [7]. If a problem is found, the batch is discarded, and the focus is on corrective actions to fix future batches. Pharmaceuti- cal QbD is a more efficient, proac- tive approach to deal with quality. QbD is a way to predict product performance based on design inputs [8]. It emphasizes that robust for- mulations and processes should be able to accommodate typical vari- ation seen in APIs, processes, and excipients without compromising the product's manufacture, stability, or performance. To understand the effect and interaction of different factors on product performance, input variables should be varied in a purposeful way [3]. Purposeful variation of excipi- ent properties relates to all typical variation that users can expect when using a specific excipient. This is in contrast to the approach where the full specification is considered as expected variation. Reliable excip- ient suppliers are often not able to supply or produce excipients that are at the edge of specifications. They produce in a narrower bandwidth than specification, aiming to be as consistent as possible. Multivariate analysis (MVA) is a key concept in the QbD approach that allows for identification of typical variation in large data sets. MVA is a set of statistical tools that can be used to evaluate relation- ships within large, complex data sets in a scientific, risk-based way. Principal component analysis (PCA) is an MVA tool that is particularly useful to investigate patterns and clusters in a large data set, even when patterns are not visible with Pauline H. M. Janssen and Bastiaan H. J. Dickhoff DFE Pharma