The role of PAT in biotechnology

Product quality of pharmaceuticals manufactured in biotechnology processes is to a large extent synonymous with the reduction and control of unwanted biological side-products. Production of biopharmaceutical proteins and secondary metabolites such as antibiotics are the result of biosynthetic capacity of the microbes or cells used. But this capacity may also contribute to turn the product molecules into molecular forms unacceptable in the final product, form side-metabolites that could harm the cell physiology and release contaminating cellular components. The main task for the ensuing process steps, the downstream procedures, is to reduce these impurities to acceptable levels that comply with the regulatory quality criteria.

The PAT initiative1-3 clearly addresses the need for developing methods that, instead of only testing quality, will also monitor on-line, assess, predict and control quality in the manufacturing process, earlier and more precisely. All this is designed to facilitate faster release and elevate manufacturing safety and reliability to higher levels.

In pharmaceutical manufacturing of chemicals, PAT methodology has leaned more towards later stages, including crystallisation, tablet analysis and final formulation. For example, techniques such as near-infrared (NIR) spectroscopy4 and Raman spectroscopy5 have provided new useful methodologies for evaluation of particle size6, crystallinity properties7 or of assessment of tablets8.

PAT for biotechnology-related pharmaceuticals manufacture should preferably be focused on the earlier steps in the manufacturing processes where the quality properties are generated and possible to control. This article outlines some of the preconditions for that.

A typical biopharmaceutical process

A typical biopharmaceutical process for production of a pharmacological active protein (see Figure 1) is centered on the bio-reaction step where a biological system makes the product protein by biosynthesis. Today, this is normally carried out in genetically modified organisms such as a bacterium or a well defined animal cell line. The biological system consumes nutrients for its own growth and the formation of metabolites used in the biosynthesis of the product protein. In the manufacturing process this takes place in a bioreactor operation. Upstream the bioreactor nutrients are prepared with accurate compositions and fed in a precisely controlled way to the bioreactor. The environment inside the bioreactor is carefully controlled in order to provide optimal reaction conditions for the organisms. Downstream of the bioreactor the product protein is purified and impurities and possible contaminants removed. This normally requires a relatively large number of selective purification steps that consecutively remove or reduce the impurities. Downstream processing is finalised by formulating the protein in a suitable liquid or powder form.

After the purification and formulation operations the purity of the product protein must meet the regulatory requirements for approval and release. This is based on extensive quality testing of the final product formulation.

Whether a pharmaceutical protein is produced in bacteria, yeast, fungi or animal cell cultures, the generalised scheme of a biopharmaceutical process as outlined in Figure 1 looks more or less the same. The basic engineering design requirements are similar but must always be adapted to the cell type of use, the medium composition, conversion rates, volumes and impurity patterns.

With biological systems that are not genetically modified, nowadays mostly for monoclonal antibodies and vaccines production, the same manufacturing principles apply. Production of proteins in eukaryotic organisms normally results in glycoproteins with varying degree of glycosylation. This gives a variation of product protein structure and molecular weight which is sometimes acceptable, sometimes not, for which the purification process must be adapted. If a small pharmaceutical molecule is produced, for example an antibiotic structure, the process scheme differs somewhat more in the downstream processing operations. For example, extraction procedures are common for purifying and concentrating the antibiotic.

Purity and impurity determination

Many of the process analytical requirements for biopharmaceutical processes are different from chemical pharmaceuticals. The biological system generates a number of impurities of biological origin. The concentrations of these impurities vary considerably, but are often low or very low. The impurities must be removed and/or reduced to defined levels. The regulatory approval of the drug has been based on detailed reports and validated methods for the pattern of impurities9,10.

Furthermore, contaminants of foreign biological species may occur since the environmental conditions in the processes are favourable for infection. These contaminants must be detected with high sensitivity.

Table 1 summarises commonly occurring impurities and contaminants in biopharmaceutical protein manufacture and how these are typically quantified in quality and process control laboratories. The host cell system generates a variety of biomolecular compounds from cellular activities and cell structures. One example is endotoxins which are analysed with the well established LAL (using Limulus amebocyte lysate) method. The host cell proteins, which should not occur in the final product formulation, are analysed with standard bioanalytical methods such as SDS-PAGE and immunoassays. The recombinant product protein is to some extent proteolytically degraded, transformed or modified by more or less known reactions. These forms of the protein are per definition impurities and must be quantitated at least in the final product. Common methods for that are, depending on structural form, HPLC, capillary electrophoresis or immunoassays.

Nucleic acids, especially foreign DNA, must be reduced to very low and quantifiable values. Occurrence of contaminating species, such as foreign microorganisms, must be detected and determined, which takes considerable time.

Ideally, these impurities should be monitored not only in the final formulation of the drug, but after each downstream step, or at least the steps aimed to reduce specific impurities. Such in-process analyses are time-consuming, costly, since they are normally done manually, and deliver data too late for using these to adjust the running process.

Process Analytical Technology

FDA has defined Process Analytical Technology as a system for designing, analysing and controlling pharmaceutical manufacturing through timely measurements of critical quality and performance properties of raw materials and in-process materials and processes with the goal to ensure final product quality1. ‘Analytical’ is defined as all chemical, physical, microbiological, mathematical and risk analysis procedures which could be conducted in an integrated manner to achieve the goal. The FDA’s leading idea is that “quality cannot be tested into the products; it should be built in or should be by design”.

This is enabled through a comprehensive understanding of:

• The intended therapeutic objectives, patient population, route of administration, pharmacological, toxicological and pharmacokinetic characteristics of a drug
• The chemical, physical and biopharmaceutical characteristics of a drug
• Design of a product and selection of product components
• The design of the manufacturing process with all required principles

With the biotechnology manufacturing view outlined above, most of these statements could also apply in a broader sense. The challenging question is: to what extent can it be efficiently realised in biopharmaceutical production?

Some examples of PAT from biotechnology-based pharmaceutical processes

So far the main focus of PAT has been on monitoring and controlling the bio-reaction part of the process. By monitoring the formation of total cell mass and the main product molecules the productivity of the process can be determined12,13. In addition to that, one often wishes to monitor the key reactants in the nutrient medium, mainly the carbon and nitrogen sources, as well as the uptake of oxygen in order to determine the consumption rate of these. Monitoring of side-products from the cellular metabolism that can inhibit the bioactivity and reduce the efficiency of the process adds further important information. Taken together, these parameters allow on-line calculation of the productivity and yield of the process.

Here, near infrared spectroscopy, often demonstrated as useful for chemical pharmaceutical PAT applications4,6, has proven to be a valuable tool14. The NIR spectra can be sampled in real-time by fiber-optical probes placed in the bioreactor with cells and nutrient media. An example of this is recombinant vaccine production (cholera toxin B protein) in a fed-batch bacterial fermentation with Vibrio cholerae where the concentrations of the glucose substrate, the acetate side-product, the bacterial mass and toxin protein were monitored on-line by NIR spectroscopy and subsequently used for controlling the substrate feeding15. Another NIR example is the at-line monitoring of an antibiotics production bioprocess where the antibiotic ingredients could be monitored in the bioreaction and downstream processing parts16. In these two examples chemometric methods have been applied to analyse the NIR spectra with the help of elaborate models allowing several constituents to be predicted simultaneously. This is a powerful strength of NIR for PAT applications in biotechnology because it leads to significant simplification of the otherwise complicated analytical handling of the crude process samples.

Another example demonstrates how the cell mass in a recombinant protein production in E. coli fed-batch fermentation can be estimated successfully with an alternative multivariate modeling method, an artificial neural network, by using existing standard process data such as O2/CO2 gas analysis and titration for pH control17. In this example it is clearly shown that a successful PAT application may not necessarily need new sensors but can manage by elaborating on data already available in the process computer software, provided a good mathematical modeling solution can be developed18.

As pointed out above, to really have impact on biopharmaceutical manufacture PAT applications need to be invented, developed and applied at the important downstream processing steps during the product purification. In particular impurities must be monitored rapidly so that there is time for correctly adjusting the effluent stream before the next processing step starts. As evident from Table 1 this requires that methods such as HPLC, SDS-PAGE, or CE are adapted to on-line and automatic operation or robotised. A demonstration of this was given by Riley et al where a HPLC method was applied on-line for pooling chromatographic column effluents19. One conclusion from this work is that if a validated off-line method already exists, it seems feasible to try to adapt this to automatic operation.

Quality versus economy

Quality and economy are tightly interconnected in pharmaceutical manufacture. Process economy is traditional considered as a classical optimisation problem. How should process and production parameters be adjusted to maximise the productivity, yield, throughput or other properties related to the capacity of the manufacturing plant? Of particular interest is the corresponding initial investments and return of capital. These are important concerns in a manufacturing company. As a consequence, process optimisation becomes a core engineering activity.

But optimisation improvements without keeping quality within the required boundaries become useless. To manage with this, simultaneous monitoring of both quality-related parameters and process economy-related parameters are necessary; in the best case at the same time.

Conclusion

A clear difference in focus is apparent when comparing PAT applications for pharmaceuticals produced in chemical and biotechnology processes. Still, several of the analytical methods coincide, such as NIR, Raman and UV spectrometry.

Furthermore, modeling methods, either based on chemometrics principles or other mathematical modeling principles, are fundamentally the same.

In the above PAT examples from different biotechnology pharma processes it is obvious that computer communication between process equipment, analytical instrumentation and powerful computation software is a necessity. Today’s industrial process monitoring control systems would benefit from a higher degree of communication flexibility in order to exploit the possibilities of new computation methods and programs20. Furthermore, for industrial applications in the pharmaceutical industry very high stability and reliability of the computer systems could not be neglected.

The interrelationship between quality and process economy parameters projected onto the previous generalised biopharmaceutical process scheme is illustrated in Figure 2. However, there is a profound need for developing new just-in-time in-process monitoring methods, sensors and prediction models; a need which should be paired with a deeper understanding of these interrelations’ underlying causes.

Moreover, if, for comprehensive process understanding, all of the four points mentioned above can be disentangled for pharma biotechnology production, a significant step forward can be achieved.

References

1. US Food and Drug Administration, Center for Drug Evaluation and Research (2004) Process analytical technology (PAT) initiative. www.fda.gov/cder/OPS/PAT.htm
2. European Medicines Agency EMEA (2005) Inspections – process analytical technology, www.emea.eu.int/Inspections/PAThome.html
3. Koch M, Hamad ML (2004) Evolving and improving PAT. Eur Pharm Rev 9, 46-48
4. Herkert T (2004) NIR, the perfect PAT tool for a solid dosage facility. Eur Pharm Rev 9, 30-34
5. Rantanen J (2005) Raman spectroscopy, a process analytical tool. Eur Pharm Rev 10, 53-58
6. O’Neil AJ, Jee RD, Moffat AC (1998) The application of multiple linear regression to the measurement of the median particle size of drugs and pharmaceutical excipients by near-infrared spectroscopy. Analyst 123, 2297-302.
7. Otsuka M, Kato F, Matsuda Y (2000) Comparative evaluation of the degree of indomethacin crystallinity by chemoinfometrical Fourier-transformed [corrected] near-infrared spectroscopy and conventional powder X-ray diffractometry. AAPS PharmSci 2(1), E9
8. Johansson J, Pettersson S, Folestad S (2005) Characterization of different laser irradiation methods for quantitative Raman tablet assessment. J Pharm Biomed Anal 39, 510-516
9. Webber K (2005) FDA update: Process analytical technology for biotechnology products. J Proc Anal Technol 2, 12-14
10. Garnick RL, Ross MJ, Baffi RA (1991) Characterization of proteins from recombinant DNA manufacture, in Drug biotechnology regulation, scientific basis and practice (Eds. Y-Y Chiu. J L Gueriguian), Marcel Dekker Inc, New York
11. Dinner A, Fose JM (1991) Industry’s experience with worldwide regulation of biotechnology products, in Drug biotechnology regulation, scientific basis and practice (Eds. Y-Y Chiu. J L Gueriguian), Marcel Dekker Inc, New York
12. Hinz DC (2005) PAT for API. Eur Pharm Rev 10, 81-86
13. Liu Y, Wang F, Lee W (2001) On-line monitoring and controlling system for fermentation processes. Biochem. Eng. J 7, 17-25.
14. Harthun S, Matischak K, Friedl P (1997) Determination of recombinant protein in animal cell culture supernatant by near-infrared spectroscopy. Anal Biochem 2511, 73-78
15. Navratil M, Norberg A, Lembrén L, Mandenius CF (2005) On-line multi-analyzer monitoring of biomass, glucose and acetate for growth rate control of a Vibrio cholerae fed-batch cultivation. J Biotechnol 115, 67–79
16. Lopes JA; Costa PF, Alves TP, Menezes JC (2004) Chemometrics in bioprocess engineering: process analytical technology (PAT) applications. Chemometr Intel Lab Sys 74, 269-275
17. Jenzsch M, Simutis R, Eisbrenner G, Stuckrath I, Lubbert A (2006) Estimation of biomass concentrations in fermentation processes for recombinant protein production.Bioprocess Biosyst Eng. Feb 25 2006
18. Galvanauskas V, R. Simutis R, A. Lübbert A (2004) Hybrid process models for process optimisation, monitoring and control. Bioprocess Biosyst Eng 26, 393–400
19. Cooley, RE (2003) Utilizing PAT to Monitor and Control Bulk Biotech Processes, Eli Lilly www.fda.gov/cder/OPS/cooley/index.htm
20. Cimander C, Bachinger T, Mandenius CF (2003) Integration of distributed multi-analyzer monitoring and control in bioprocessing based on a real-time expert system. J Biotechnol 103, 237-48