At the beginning of this millennium, the FDA decided to allow changes to process monitoring methods on an incremental basis if the science is compelling and an improvement of product quality or decrease in the cost of manufacture (at constant quality) is the goal. This opens the door for several technologies to break through to on-line or at-line monitoring. Most of the early advantage went to the spectroscopy instruments and, as such, most of the pharmaceutical R&D on instrumentation has been focused on near IR mid IR and Raman over the last decade. There is significant experience (30 years) in the chemical and petroleum industries, if the pharmaceutical companies choose to tap it. Following on, at a slower deployment level, are chromatographic systems (GC and HPLC) to cover the evaluation of concentration and performance characteristics where spectroscopy is not effective at the detection level required. The key consideration is that these are multivariate instruments and the data they generate do not fit into the traditional world of DCS and plant historians.
This drive to multivariate instruments leads to:
- Requiring chemometric processing to automate the extraction of information;
- Borrowing from laboratory environments to handle the 3-order-of-mangnitude increase in the data flow;
- Understanding that process content has a multivariate flavor as well, leading to the need to process and display inferentials and multivariate statistical process control; and
- Addressing the complications associated with deployment of complex, expert-driven new technology.
Pharmaceutical Process Analytical Technology ties the analytical instrument more tightly to the application. As a result, performance of an instrument (like a NIR) will not rely as much on the hardware specifications or the ability to evaluate unrelated standards. Instead performance will be judged based on the suitability for a particular application. We need to show effectiveness with samples identical or very similar to those specific to an individual process. This puts pressure on applications support for any instrument that has not seen use in a pharmaceutical R&D or pilot plant environment. It also reinforces the need for functional, reliable and flexible software to manage the tasks ahead.
This instrument focus was a fundamental change for the pharmaceutical companies, for the FDA, and for process instrument suppliers. An understanding of the chemometrics role in processing the information content of instrument and process data is critical to efficient feedback for manufacturing and Infometrix is in the vanguard of this technology.