By C. Patrick McAtee, Senior Biochemist, Sachem, Inc.

Quality by Design (QbD) is a systematic process to build quality into a product from inception to final output, with the goal to more fully understand how product and process attributes relate to the product performance. The current challenge in QbD is in defining when enough is truly enough. QbD doesn't advocate any particular experimental approach, only that the approach used can be scientifically justified.

QbD works hand in hand with design of experimentation (DoE), which requires assumptions to be made about the interaction of variables and the linearity of responses. Those linearity assumptions can later be tested. DoE may not be applicable for truly novel processes, but can be a very efficient way of reducing large numbers of experiments to a more manageable level.

One could potentially describe QbD in biopharmaceutical characterization as the aggressive use of analytical methods to characterize primary, secondary, tertiary, etc., pathways in a process. In this definition, by using QbD we seek to expose, identify, quantify and evaluate as many heterogeneities, reaction by-products, trace impurities, or other artifacts as possible. The analyte-of-interest may be, upon first glance to the process scientist, the “least interesting” substance in the mixture. Yet, for example, downstream disciplines such as Drug Metabolism and Pharmokinetics (DMPK) and adsorption, distribution, metabolism, and excretion (ADME) may be ultimately affected by these “least interesting” components.

How to address these issues? In a lot of respects, QbD is simply an extension of validation—that is, providing clear evidence that a process is understood, under control, and gives the same results time after time in a robust and reproducible manner. Any subtleties that could either enhance or derail a candidate need to be identified early on but also correlated with downstream observations.

A New Brand of Proteomics

Although there are many analytical tools which contribute to this knowledge base, the ability to derive proteomic fingerprints of process streams represents a particularly powerful approach to address quality concerns, particularly in reference to trace impurities, which can be identified and evaluated for further investigation. It is true that “classical proteomics” is typically measuring tens, hundreds or thousands of variables per entity. In recent times, proteomics and the tools of proteomics, in particular, have evolved into more sophisticated and targeted approaches to address specific drug discovery and manufacturing issues.

In addition to reproducibly identifying a peptide component in a particular sample, proteomics offers the ability to identify key posttranslational events as well. However, proteomic approaches can address the variability of product isoforms in bulk drug substance, between repeat bioreactor runs, variance with cell line/bank changes, changes in growth medium, process scale up, tech transfer and downstream processing.

“Classical proteomics”, centered primarily on profiling, has been dependent upon the use of two-dimensional gel electrophoresis and MALDI-TOF. Instrumentation and software have evolved. Electrospray instrumentation has largely displaced MALDI-TOF due to its ability to generate specific sequence and posttranslational information. Electrospray mass spectrometry coupled with HPLC is a very powerful approach for analysis of peptides yet the complexity of proteomic samples typically necessitates a two-dimensional approach. The most frequently used approach for two-dimensional separations of peptides by HPLC has been the use of strong cation exchange in the first dimension followed by reversed phase in the second dimension.

While there are numerous studies that show the value of 2-D HPLC in analysis of complex protein mixtures, I propose that proteomics can play a central role in unifying data in the pharmaceutical process and hence providing an interactive fingerprint for the development of an efficacious preparation. How would one handle the throughput and correlation analysis in such an undertaking? The advancement of software for data formulation coupled with automated UPLC systems enables the processing of this complex information in a timely and reproducible manner.

Proteomic approaches readily identify large numbers of proteins and their postranslational modifications, but the value of proteomics as applied to profiling clinical materials would allow for identification and tracking of clinically relevant proteins in response and resistance.

The preferred sample source for clinical tracking is typically blood plasma. Yet broad use of this protocol is challenged by the fact that the primary blood proteins may mask the detection of less abundant proteins of interest, which constitute the majority of useful clinical markers. Several procedures have been designed to remove these more abundant proteins before proteomic analysis. However, these methods may sacrifice other proteins by nonspecific binding, thus lowering the screening efficiency.

Displacement Chromatography

The introduction of displacement chromatography as the first dimension followed by typical reversed phase UPLC in the second dimension overcomes this barrier. Displacement chromatography exploits the nonlinear, competition between the components to be separated, resulting in higher resolution, particularly among closely related species. In contrast, in elution processes (e.g., linear gradient, step gradient), the separation takes place under relatively weaker binding conditions (which are essential elute the components off the column). The separation factors among solutes are thus lower in elution than in displacement, leading to poorer resolution in the elution modes of operation. Other advantages of displacement include a better control over the product concentrations and the emergence of the product in relatively low concentrations of the mobile phase modifier. Coupled with UPLC, this will result in a combination of high throughput and high resolution in a single process.

Two-dimensional chromatography represents the most thorough and rigorous approach to evaluation of the proteome. While previously accepted approaches have utilized elution mode chromatographic approaches such as cation exchange to reversed phase HPLC, yields are typically very low requiring analytical sensitivities in the picomolar to femtomolar range. As Displacement chromatography offers the advantage of concentration of trace components, two dimensional chromatography utilizing displacement rather than elution mode in the upstream chromatography step enables a powerful process for analysis of trace components, modifications, and identification of minor expressed components of the proteome.

This 2-D proteomic approach can be applied in a variety of biopharmaceutical development applications such as a more detailed characterization of the product during upstream process development,  to identify differences due to titer increases or other process modifications before they become downstream challenges.

Comparison of a product’s 2-D "fingerprint" from one manufacturing step to the next allows the direct measurement of the effectiveness of that process step can be made. However, fingerprints can also be applicable to early and late stage clinical information in that they offer the potential to align both upstream and downstream variation effects. This is in addition to the correlative effects that downstream and upstream have on each other in biopharmaceutical manufacturing.
This approach expands upon and refines the criteria for QbD recently established by the FDA in that the product and process performance characteristics are scientifically designed to meet specific objectives rather than being empirically derived from performance of test batches. The incorporation of proteomics into the manufacturing process yields a dynamic system in which input and endpoint responses may be correlated with confidence. Simply put, if variability and the effects of variability necessitate a QbD approach to pharmaceutical manufacturing, then proteomics provides a well tested roadmap for this approach. The incorporation of displacement into 2-D UPLC fingerprints brings a newer, more complete picture that can truly be enabling.

About the Author

Dr. C. Patrick "Pat" McAtee has has been aptly labeled as a scientist who takes a commando approach to research and inquiry. His work includes the development of an efficacious vaccine to Hepatitis E, the discovery of the Hepatitis G virus as well as the deorphaning of disease-related kinases and peptidases. He is the author of numerous publications and patents in the areas of  applied genomic technologies and infectious diseases. He received his PhD in Biochemistry from The University of Chicago Pritzker School of Medicine and also completed postgraduate programs at Massachusetts Institute of Technology and The Harvard School of Public Health. Dr. McAtee has also been very active in coaching girls basketball and is the founder and head coach of the AAU Texas Shooting Stars Girls Basketball Team.