Biomedical Engineering Reference
In-Depth Information
TABL E 11.1. Examples of Analytical Methods That Could Be Considered
to Provide High False-Positive or False-Negative Rates a
“Too Sensitive” High
False Positives
“Too Insensitive”
High False Negatives
NMR Visual Observation
LC-MS Compendia b
ICP-MS Osmolality
LC-DAD pH
a A false-positive result would discover a difference between raw material lots that do
not have an impact on production; a false negativewouldmean materials would pass the
analytical test then impact production.
b Tests described in one or more of the following compendia USP, EU, EP, or ACS, and
so on.
NMR, mass spectroscopy, HPLC, and ICP-MS are not commonly considered for
routine use in rawmaterial analysis. Each method depends on the consistent operation of
very delicate and sensitive instruments. Part of the example presented below involved
developing each analytical tool to see if it was predictive, robust, and rugged enough for
raw material applications. The strategy used to develop each method included data
preprocessing and statistical model development and was based on current advances in
methods used for metabolomic applications.
Each of the analytical tests evaluated had the potential to become a platform
technique for routine analysis of complex mixtures of raw materials in water. The
techniques adapted were mass spectroscopy, NMR, and various types of HPLC. In each
case, the idea was the same: take a specific, repeatable technique and interpret the results
with pattern recognition software. Unique data pretreatment was evaluated for each type
of data, then standard MVAwas applied to help interpret the results. Details about these
analytical tools are available elsewhere [5-17] and will be discussed here only to
illustrate how QbD for complex rawmaterials encompasses not only analytical methods
and multivariate analysis, but more importantly, a proactive strategy that helps ensure
that quality is designed into a manufacturing process even when the degree of future
variation cannot be predicted.
11.7.1 Hydrolysate Example
To illustrate the concepts for QbD applied to raw materials, consider an example
involving a natural component added to media in a commercial biologic manufacturing
process. Out of trend results for the bioreactor coincided with a change in the lot of
hydrolysate. Small-scale bioreactor experiments confirmed that lot-to-lot differences of
this rawmaterial accounted for the change. An investigation ensued with two objectives:
to devise an analytical test that distinguishes acceptable and unacceptable raw material
lots and to identify specific components in the natural product that might account for the
change. Although this was an investigation and not proactive process design, this
 
Search WWH ::




Custom Search