How can you effectively validate bioseparation methods?

This insightful article underscores the critical importance of meticulously defining the purpose and scope in validating bioseparation methods. By identifying target products, contaminants, and quality attributes, it provides a comprehensive guide for establishing clear objectives, boundaries, and assumptions in the validation process. A must-read for professionals seeking precision and excellence in bioseparation methodologies.

Customer requirements. Customer requirements. Customer requirements. I can't say this enough. What do the downstream customers (internal and external) really want? Ask them! Make that a "Critical to Quality" requirement for process/product development. Does the downstream customer require a certain purity grade? Maximum particulate content? Viscosity?? Ask the customer (internal customer = inter-department handoffs, external customer = source vendors, retailers, etc). Make a list, make those requirements your objectives. Get real-world data from real-world requirements. I cannot seem to stop ranting about this.

Well, for bioseparation, depending on the downstream application, accuracy, precision, etc. are more important for the requirements of an analytical system, (e.g. HPLC-UV or LIF or CAD detector, LC-MS/MS, GC, etc.). This is because analytical detection systems are inherently incapable of further discrimination of different chemicals except for those with specific properties (e.g. fluorescence, UV-Vis absorption, Ionization-to-Mass detection, etc.). If bioseparation is for preparative purposes, e.g. biopharma applications, protein/peptide/lipid/nucleic acid/carbohydrate production, then analytical requirements are not so important. To distinguish the requirements of analytical requirements is the essence of analytical chemistry.

Defining the acceptance criteria and predetermining the acceptable variation between runs is key. +/- 3sd analysis determines the analytical variation between runs. System suitability analysis helps determine the suitability of the system for the test under performance. Incorporating qc samples and calibration samples between test samples help in reducing analytical variation and also in accurately quantifying the test sample. Incorporating internal standard and running reference std helps in accurate identification. Baselines must be consistently set, indicating the pressure within the system is OK. System drift over time can be handled by repeating re-validation/re-qualification at predetermined intervals.

Yes, all this is very important. Imagine that human, random, or systematic error/bias is introduced during validation. What would happen if the method was then transferred into production? Could production scientists possibly achieve consistent or reproducible results? Likely not. During validation, the technical skill requirements, stability of the process, must be thoroughly vetted for the purpose of routine analysis by a person who is skilled but is not a demi-god of chemistry. Meaning, the method must be ironed-out of kinks such that the average scientist can perform the method routinely. Otherwise, you will have an unending stream of quality problems later on and many unhappy, frustrated, and angry scientists. Make people happy.

See my previous comment above regarding the consequences of letting human/random/systematic error and/or bias pass through the validation process. Dire consequences and a slew of curses from colleagues in production will only ensue. Beware!

From my experience at PPD (Thermo Fisher's CRO branch), documentation and even method writing was part of a boiler-plate process. When work like formatting, organizing raw data into clear, concrete, human- and machine-readable databases is practically automated and turned into routine work, the real knowledge work that is required of scientists is liberated. We have to set scientists free. Free from what? Mundane and easily avoidable problems, like typos, clerical errors, wrong data in wrong location, dumb issues like that. The real work and the real problems that scientists should face ought to be abstract and scientific, not clerical. Everyone hates routine documentation, believe me. Make this step the easiest and most pleasurable.