All4Nutra
Abstract
The integration of Process Analytical Technology (PAT) into pharmaceutical process development has become a critical focus since the issuance of FDA PAT guidance in 2004. While PAT applications for blend uniformity (BU) and content uniformity (CU) are well-established, their utilization in early-phase development in an end-to-end fashion is seldomly reported. This study presents a novel case of deploying end-to-end PAT capabilities, combined with material-sparing chemometric approaches, to mitigate the impact of coarse API particle size distribution on BU and CU. Near-infrared spectroscopy (NIRS) data were collected across three interfaces—bin-blender, tablet press feedframe, and tablets—to generate high-density real-time process data. Chemometric modeling via Classical Least Squares (CLS) was employed to translate spectra into actionable concentration data without the need for additional API calibration samples. Furthermore, this study provides a comparative analysis of CLS based upon measured versus estimated pure component spectra, highlighting the advantages and limitations of each approach. The findings underscore the potential of material-sparing PAT methods to enhance process understanding and robustness in early-phase development. The results also set the stage for broader adoption of PAT and phase-appropriate method development, fostering innovation in pharmaceutical research and development.
Introduction
Since the issuance of process analytical technology (PAT) guidance by FDA in 2004, the use of PAT has been permeating many aspects of pharmaceutical process development. In drug products, the use of PAT to provide real-time monitoring on both blend uniformity (BU) and content uniformity (CU) have been two most commonly-reported applications, given their criticality for both process development and commercial manufacturing. Given the accumulated experience in these areas on the field, Innovation & Quality (IQ) consortium recently published a white paper (Bautista et al., 2025) to propose a paradigm shift from traditional sampling to determine BU and CU to the combinational use of risk assessment and PAT to ensure BU and CU in pharmaceutical oral solid dosage forms, including tablet and capsules.
Literature survey suggests three-folded intended purposes of PAT commensurate with the progression of pharmaceutical process development, including process understanding, in-process control (IPC) and real-time release testing (RTRt). The use of PAT for IPC and/or RTRt is often more easily justifiable from a monetary impact standpoint than its use for process understanding given their relevance to GMP manufacturing. That being said, the value proposition of PAT for process understanding is often underrated given it is difficult to be quantified by a monetary impact. Especially for early-phased development when both API supply and process understanding are limited, if at-scale PAT tools can be deployed, the data collected can not only provide insights into scale- and phase-dependent process variability, but also enhance late-phase process development by improving the robustness of PAT tools designed for IPC or RTRt.
The material and resource intensive requirement to calibrate a PAT sensor is commonly acknowledged on the field, especially for early-phased pharmaceutical development when API supply is often a luxury to have. Facing such a challenge, pharmaceutical researchers have been adopting innovative approaches to minimize the calibration requirement during process development, including the use of scale-down equipment to reduce the consumption of representative calibration samples(Hetrick et al., 2017, Li et al., 2018) and the use of material-sparing chemometric approaches to predict API content via the mere use of pure component spectra (Muteki et al., 2013, Shi et al., 2018). Two mainstream algorithms applied so far for pharmaceutical applications are classical least squares (CLS, (Bondi et al., 2012a, Bondi et al., 2012c, Shi et al., 2012b) and iterative optimization technique (IOT, Henson et al., 2024; Rish et al., 2022a; Rish et al., 2022b; Rish et al., 2023; Alam et al., 2021; Harms et al., 2019). The two algorithms mainly differ on its mathematical constructs, from which CLS leverages closed-form analytical equations to estimate chemical concentrations, and IOT leverages optimization-based routine with constraints for the same purpose. Despite such a difference, both algorithms rely on either measured or estimated pure component spectra to estimate chemical compositions. Most applications of these two algorithms have been reported on pharmaceutical powder mixture (Garcia-Munoz and Torres, 2020, Shi et al., 2023). The use of these material-sparing approaches on determining API content in both blends and tablets have not been reported. Additionally, no studies have been reported so far on the advantages and disadvantages of using measured vs. estimated pure component spectra to predict chemical composition of mixture spectra via either CLS or IOT.
In this paper, we report a case study in early-phased process development, demonstrating the use of end-to-end PAT capabilities in our lab from powder blending to tablet compression and to stratified core tablets, combined with material-sparing chemometric approaches. This integrated methodology enables the collection of high-density real-time data, thus mitigating the risk associated with the coarse API particle size distribution (PSD) on drug product content uniformity. Real-time near infrared spectroscopy (NIRS) data were collected on two conformance batches (i.e., one batch for each dose strength) of a clinical asset via three different PAT interfaces, including powder mixing in a bin-blender, in the feedframe of a tablet press and tablets themselves. Interface-dependent material-sparing chemometric approaches (i.e., CLS) were used to translate real-time spectra to concentration information without additional use of APIs to prepare calibration samples. The relatively low dose of the API (i.e., 5 %, w/w) also presents opportunities to conduct a comparison between the use of measured and estimated pure component spectra for CLS. It is our intention to showcase the potential of material-sparing chemometric approaches in early-phased development to improve process understanding and to encourage the broader adoption of PAT starting from early-phased formulation and process development. Meantime, it is also our intention to provide guidance on how to best utilize the material-sparing nature of CLS/IOT for phase-appropriate PAT method development across different formulation compositions. Given the similarity between CLS and IOT reported so far, it is expected that the learnings on the pros and the cons of applying measured vs. estimated pure component spectra is equally applicable for IOT as well.
Read more here
Zhenqi Shi, Prajwal Thool, Madisen Omstead, Cheng Chiang, Nivedita Shetty, Deepak Prasad, Helen Hou, Chen Mao,
A case study on the use of end-to-end real-time monitoring to ensure quality of oral solid dosage forms in pharmaceutical development, International Journal of Pharmaceutics, 2025, 125775, ISSN 0378-5173, https://doi.org/10.1016/j.ijpharm.2025.125775.
Read also our introduction article on Orally Disintegrating Tablets (ODTs) here:
