Using NIR to Move Bioprocessing into a PAT Framework

Process analytical technology (PAT) is a well-publicized initiative championed by the US Food and Drug Administration to promote innovation in pharmaceutical processing (1). A pharmaceutical manufacturer following PAT guidelines will be able to understand areas where process variability may occur and respond instantaneously to account for that variability. This responsiveness results in higher product quality and reduced cost due to lost batches. In addition to the ultimate goal of better process understanding, FDA's guidance for industry, A Framework for Innovative Pharmaceutical Development, Manufacturing and Quality Assurance, enumerates the potential benefits from gains in quality, safety and efficiency, including (1):

• Reducing production cycle time by using on-, in- or at-line measurement and controls

• Preventing rejects, scrap, and reprocessing

• Real-time release

• Increasing automation to improve operator safety and reduce human errors

• Improving energy and material use and increasing capacity

• Facilitating continuous processing to improve efficiency and manage variability (1–3).

Despite these advantages, processing using PAT in the pharmaceutical industry has lagged behind other industries such as the semiconductor-manufacturing or food-processing industries. PAT initiatives in the biopharmaceutical industry have been even slower (4).

Many aspects are involved in a successful PAT process. Cultural acceptance within the corporate structure ultimately drives the success, but having confidence in the instruments and hardware can also drive acceptance. The key physical components of a PAT-compliant process are the analytical probes and instruments used to monitor and control the process. A variety of instruments are available to the pharmaceutical industry, including simple temperature and pH probes, oxygen and total organic carbon probes. More complex on-line high-performance liquid chromatographs (HPLC) and mass spectrometers are available as well, but one of the most commonly relied upon and robust techniques is near infrared (NIR) spectroscopy. NIR systems are of primary importance in PAT-compliant processes because they do not destroy the sample during analysis. In addition, NIR systems can be placed at, in, or on-line, and achieve results automatically and in real time.

NIR as part of PAT

Unlike mid-infrared spectroscopy, the material under NIR analysis usually does not need to be diluted or manipulated robotically or by a laboratory technician to achieve results. The technique allows direct sampling of many materials, enabling probes to be placed directly in the process stream. The NIR light is of the same type used in the telecommunications industry. It is easily shunted from the probe to the instrument over long distances using commercially available fiber-optic cables. In addition, Fourier transform (FT) instruments provide precise and reproducible results—normally in just a few seconds. Best results are achieved using NIR systems that have been originally designed for process streams. These systems communicate easily and directly with process-control systems.

NIR spectroscopy relies on the interaction of light to analyze a variety of raw materials, mixtures, intermediates, and finished products. Specific photons of light with frequencies between 12,000 and 4000 cm^-1 can be absorbed by different chemical bonds, which set up characteristic vibrations within the molecules. Light that is not absorbed by the bonds is collected and displayed as a spectrum. The inherent complexity of mixtures and the unavoidable interactions between vibrating molecules can make interpreting NIR spectra very difficult. The advent of chemometrics and powerful computing algorithms, however, has greatly simplified interpretation.

NIR analysis usually requires training the system to recognize concentrations of analytes or the identity of materials. A series of standards of known composition are characterized with a primary method and scanned with the NIR system. The key to a robust NIR analysis is to have adequate and appropriate standards for this training, as well as accurate primary data on the composition of the material. The spectroscopic and composition data are deconstructed through chemometric software, which can correlate even subtle spectroscopic variation with the material composition, and this method can then be used to determine the composition of subsequent samples. In addition, several components within a mixture can be determined simultaneously from just one spectrum.

The power of NIR in a PAT environment is that it provides crucial comprehensive information about a process in real time while the material is being manufactured. Other spectroscopic, electrochemical, chromatographic, or wet-chemistry techniques require samples to be drawn off and sent to the laboratory for analysis, which, in addition to the cost in labor and time, necessitate the destruction of valuable samples. Analysis of multiple components in a mixture also requires several analytical techniques that invariably add to the cost. Most importantly, the information obtained by these off-line techniques may be several hours or days old, which is too long to maintain adequate control over the process.

Once in place, NIR largely replaces these off-line analytical techniques, and the speed with which NIR can nondestructively determine multiple components in a mixture allows critical decisions and adjustments to be made before the material falls out of specification and becomes unsalvageable. Batches of active pharmaceuticals may easily be valued in excess of $1 million, which is an expensive loss if out-of-specification material is discovered during final quality checks simply because the process was not tightly controlled using PAT.

Bioprocessing

Biological materials are an increasingly important field in the pharmaceutical industry. There are currently more than 350 biologicals approved by FDA (5). Of these, 56 are vaccines, 130 are antibodies, and 30 are labeled as recombinant. Bioprocesses that produce pharmaceuticals rely on intricate biological systems to synthesize useful products, many of which are large complex proteins, hormones, or polysaccharides that are difficult, or even impossible, to manufacture in large quantities any other way. The complexity of these biological systems and the finished product makes variability among batches inevitable. In fact, it is estimated that 30% of production batches require reprocessing for quality reasons and, although some of these reprocessed batches continue through to production, this additional process, along with the discarding of poor quality product, can result in an approximately 10-fold loss in profit (2). Therefore, if there ever was an industry in need of PAT control, the biopharmaceutical industry is it.

Many types of cells, including genetically engineered bacterial and yeast cells are used to culture and produce biopharmaceutical products. The majority of products, however, are produced within mammalian systems such as Chinese hamster ovary (CHO), green monkey (VERO), or human embryonic kidney (HEK) cell lines. These mammalian-cell lines require more controlled feeding and care than other lines because they originate from multicellular organisms that have been adapted to survive as free-floating cells in a synthetic or complex artificial medium. The protein products are well-characterized biologics that require precise control of the culture parameters. Often, proper glycosylation of the target protein is important to its function, and post-translational modifications such as glycosylation are dependent on subtle glutamine feeding as well as the control and removal of ammonia waste products (2). Specifically, the ratio of N-acetyl-neuraminic acid (NeuAc) to N-glycolyl-neuraminic acid (NeuGc) of the target protein needs to be maximized to optimize the in vivo half-life of the biopharmaceutical. A major influence on this ratio is the composition of the cell culture medium (6). Controlling the cell culture medium through NIR and other PAT instruments helps ensure the correct NeuAc/NeuGc ratio.

Figure 1 represents a typical industrial-scale bioprocess system producing and purifying a target complex molecule. Raw materials (see Figure 1A) used in media preparation and culture feeding are mixed and diluted to appropriate concentrations. Inoculums containing the carefully produced and stored cell lines are grown in small containers (see Figure 1B) prior to introduction into the large-scale process tank (see Figure 1C). Depending on the process, the cells are nurtured in the tank for several days where they grow and produce the target molecule. Typically, for mammalian cells, the target molecule is a protein that is secreted out into the cell medium where it accumulates, but other expression systems such as bacterial cells, may produce and store the target molecule within inclusion bodies and require eventual break-up of the cell for recovery. After a prescribed time, the spent medium and products are withdrawn from the process tank and purified, which may involve microfiltering or centrifugation (see Figure 1D). For mammalian cells, this step is a 'soft' method so as not to disrupt the cells, as cellular debris is more difficult to remove from the target protein. The resulting material is typically concentrated approximately 10-fold by tangential flow filtration (see Figure 1E). Often, diafiltration further purifies the filtrate and places it in a binding buffer that is suitable for the subsequent chromatography step. Large low-pressure chromatography columns (see Figure 1F) containing a binding resin (frequently, Protein A in the case of antibody purification) selectively bind the target molecule and bring it out of solution. Following wash steps that are optimized for impurity removal, the columns are flushed with an elution buffer allowing the product molecule to be released and collected. Finally, polishing steps (see Figure 1G) may occur where ion and buffer exchange is performed. Otherwise, chromatography is used to separate and collect the different possible glyocoforms. This step is followed by sterile filtration, liquid fill, and lyophilization.

Figure 1: Process schematic for manufacturing biopharmaceuticals. Raw materials (A) and inoculating cell cultures (B) are nurtured for several days in a process tank (C). The resulting target molecule is purified from the spent media by centrifugation (D) and filtration (E). The target molecule is further purified with an affinity column (F) or chromatography (G) and lyophilized. Near infrared spectroscopic analysis can be performed automatically in real-time at all stages of the process.

NIR in a PAT-compliant bioprocess

There are several places within this process stream that are conducive to query and analysis by PAT instrumentation and, specifically, NIR equipment. The monitored variables are described as quality variables or manipulated variables. Quality variables describe the state of the process at a given point (e.g., concentration of the target protein), whereas manipulated variables are used to control the process (e.g., glutamine concentration) (7). NIR is easily used in conjunction with both types, and is also commonly used in raw material identification and analysis (7, 8). In addition, quality control of the process-ready medium before use is important while quality checks of the inoculums eliminate suspect samples or select the best candidates that lead to improved batch performance and reduced variability (7).

During the cell growth stage and production of the target molecule, several variables related to nutrients, waste, and overall health (e.g., glucose, glutamine, lactate, ammonia, pH, dissolved oxygen, and cell density) are monitored. In non-PAT processes, many of these components are monitored off-line using sophisticated and expensive bioanalytical systems such as a Nova Biomedical BioProfile 100 Plus analyzer (Waltham, MA) or a YSI 7100 MBS (YSI Life Sciences, Yellow Springs, OH). These analyzers are electrochemical and typically use immobilized enzyme membranes in the analysis. Considerable work and cost is required, however, to maintain consistent calibration because the enzymes degrade with time, which is a critical issue for monitoring processes that can last several weeks. In addition to destroying valuable samples in the analysis, these electrochemical analyzers require purchase and replacement of reagents, electrodes, and membranes. As a result, high annual cost for consumables can easily climb into the tens of thousands of dollars.

In a PAT process, however, many of these same components are monitored and controlled in real time with NIR analyzers coupled to process-control systems (9, 10). Figure 2 shows the results of monitoring some of these cell-culture components during a multiple day batch run. NIR is also used during the filtration, purification, and finishing steps to ensure the product has been properly concentrated and handled. Finally, the end products are often lyophilized with concurrent monitoring of the moisture content. NIR is supremely suited for controlling the lyophilization process and determining when the optimal moisture content has been achieved (7, 11, 12).

Figure 2: Plots of multiple component concentrations within a growing cell culture. All data were collected with an FT-NIR spectrophotometer (Antaris model, Thermo Fisher Scientific, Madison, WI) in-line in real time. Blue diamonds represent data collected from the spectrophotometer; red squares represent primary analytical data taken from thieved samples for comparison.

Conclusions

PAT is a useful paradigm and is making inroads into the biopharmaceutical industry, but the speed of acceptance has lagged behind other industries, including typical pharmaceuticals. This delay in acceptance is unfortunate because PAT initiatives, which are designed to develop a deeper understanding of the process and to control variability, could benefit biopharmaceuticals more than any other industry because biological systems are so inherently variable. NIR and other analytical tools required for this process understanding and variability control are already in place in other industries and will fit very well in all stages of biopharmaceutical manufacture. FDA encourages the use of these tools in a well-designed PAT process, and their use will reduce the time of manufacture and costs of reprocessed or out-of-specification product.

Todd Strother, PhD, is an applications scientist in the Scientific Instruments Division, Analyzer Group, at Thermo Fisher Scientific, 5225-4 Verona Road, Madison WI 53711, tel. 608.276.5626, todd.strother@thermofisher.com

References

1. FDA, Guidance for Industry PAT—A Framework for Innovative Pharmaceutical Development, Manufacturing and Quality Assurance (Rockville, MD, Sept. 2004), www.fda.gov/cder/guidance/6419fnl.pdf.

2. H. Trautman, "PAT in Bioprocessing," BioWorld EUROPE, 6–9 (2005).

3. J. Ganguly and G. Vogel, "Process Analytical Technology (PAT) and Scalable Automation for Bioprocess Control and Monitoring—A Case Study," Pharm. Engin., 26 (1), 1–9 (2006).

4. "Slow Adoption of PAT for Bioprocessing," Gen. Eng. & Biotechnol. News, 26 (16), (Sept. 15, 2006).

5. FDA, Center for Biologics Evaluation and Research Current Licensed Establishments and Products (Rockville, MD), Feb. 28, 2009, www.fda.gov/cber/ep/part3.htm, accessed April 8, 2009.

6. D.L. Fernandes, "Biopharmaceutical Sialylation," Euro. Biopharm. Rev., Spring, 100–104 (2006).

7. J.A. Lopes et al., "Chemometrics in Bioprocess Engineering: Process Analytical Technology (PAT) Applications," Chemomet. and Intell. Lab. Systems, 74 (2), 269–275 (2004).

8. Y. Ozaki and T. Amari, "Near-Infrared Spectroscopy in Chemical Process Analysis," in Spectroscopy in Process Analysis, J. Chalmers, Ed. (Sheffield Academic Press, 2000), pp. 53–92.

9. C. Card et al., "Near-Infrared Spectroscopy for Rapid, Simultaneous Monitoring of Multiple Components in Mammalian Cell Culture," BioProcess Inter., 6 (3), 58–67 (2008).

10. P. Roychoudhury et al., "Multiplexing Fibre Optic Near Infrared (NIR) Spectroscopy as an Emerging Technology to Monitor Industrial Bioproceses," Analytica Chimica Acta, 590 (1), 110–117 (2007).

11. J. Hirsch, "Near-Infrared Analysis of Critical Parameters in Lyophilized Materials," BioPharm Inter., 19 (2), 32–36 (2006).

12. T.P. Lin and C.C. Hsu, "Determination of Residual Moisture in Lyophilized Protein Pharmaceuticals using a Rapid and Non-invasive Method: Near Infrared Spectroscopy," J. Pharm. Sci. Technol. 56 (4), 196–205 (2002).