By Satish Mallya, Benjamin Lay, Lihong McAleer, Alexandria Emory, Nataliya Afonina
Abstract
Cell-culture media are essential raw materials that are required for the manufacture of biotherapeutic proteins. Cell-culture media are composed of multiple components, and therefore, it is difficult to develop specific identification tests for media as required by Code of Federal Regulations (CFR) 211.84. The developed testing algorithm, which incorporates a combination of few relatively simple analytical methods such as osmolality, quantitation of glucose, and folic acid, provides specific identity confirmation for seven cell-culture media with essentially similar composition that were examined in this study. The methods are well suited for routine use in the quality-control environment and the provided identification approach meets CFR, FDA, and other regulatory agencies requirements. As part of this approach, a platform glucose method that utilizes a linear standard curve was validated using alternative cell-culture media, and all established acceptance criteria were met. The assay was shown to be specific for the detection of glucose in all studied multi-component media without matrix interference regardless of glucose concentration or the vendor.
In the past few years, there has been an increase in the number of therapeutic proteins in development and those that have received approval from regulatory agencies. Proteins are produced by gene expression in bacterial or mammalian cell culture. Cell-culture media are composed of essential raw materials that are required for cell growth and expression. Similar cell-culture media may be used simultaneously in one or few biological processes at a drug product manufacturing facility. As per Code of Federal Regulations (CFR) 211. 84, each media needs to be specifically identified from other media during release testing prior to use in production (1, 2).
Cell-culture media contain multiple components such as hormones, vitamins, folic acid, lipids, amino acids, sugars, insulin, among others. Therefore, development of a specific identity test presents a challenge due to matrix interference (3). Several spectroscopic non-destructive techniques, including Fourier Transform Infrared (FT–IR), near-infrared (NIR), and Raman spectroscopy have been used for the identification and characterization of cell-culture media. FT–IR spectroscopy (mid-IR region) differentiates materials based on their chemical composition and thus, provides specific chemical identity (4). This test compares the spectrum of each media with the spectrum of each corresponding control and was successfully employed for identification of different types of growth media with diverse formulations used in vaccine production (5). However, this approach cannot be easily employed for media used in the manufacture of biologics, which might differ only by one additional component or by a concentration of similar components.
NIR spectroscopy has a light absorption much weaker in intensity compared with that of FT–IR and is based on the overtones of major bands produced in the mid-IR region. Therefore, NIR does not provide specific identity of the studied material and requires a use of spectral libraries or multivariate data analysis (MVDA) for spectra evaluation. NIR spectroscopy is a very popular method of analyzing solid and liquid raw materials because it is sensitive to matrix modifications such as surface area, sample morphology, and other sample properties (6, 7). NIR coupled with chemometric analysis was used for the characterization of soy raw material to evaluate the variability and impact on product quality. The analysis revealed that near-infrared spectra of different soy lots contain enough physicochemical information about soy hydrolysates to allow the identification of lot-to-lot variability as well as vendor-to-vendor differences (8). A few publications related to NIR spectroscopy, however, are associated with identification of complex raw materials such as cell-culture media. A combined approach of NIR and MVDA was used for the identification and qualification of basal medium powder components (9) and was evaluated as an identity tool for cell-culture media (10).
Raman vibrational spectroscopy provides a unique chemical fingerprint of molecules similar to that of FT–IR and is being used for raw materials identification, characterization, and quantitation (10, 11). However, as a non-specific identification test, it requires the use of chemometric analysis for unambiguous sample identification. Raman spectroscopy coupled with principle component analysis was successfully applied as a single integrated method for rapid identification and characterization of five different chemically defined components of cell-culture media used in a Chinese hamster ovary (CHO) cell manufacturing process for recombinant proteins (11). Although, both NIR and Raman spectroscopy are fast and non-destructive, they require the use of chemometric analysis for specific identification of cell-culture media or media components, which is not easily employed in the quality control (QC) environment. Therefore, this approach (NIR and Raman spectroscopy coupled with chemometrics analysis) is more useful for the identification of media components and screening consistency and characterization of media that are not in the QC environment.
The micellar electrokinetic capillary chromatography (MEKC) was used an alternative approach to spectroscopic methods for media identification. MEKC was successfully validated and employed for the identification of cell-culture media from Invitrogen (12). It was also used for simultaneous determination of media components such as folic, mycophenolic, nicotinic acids, hypoxanthine, and other components in protein-containing matrices from a monoclonal antibody manufacturing process (13). However, due to the complexity of sample analysis, this approach is more suitable for media characterization or in-process monitoring rather than a use as an identity test (ID) in a QC environment.
Another practical approach for media identification is the use of relatively simple tests for the identification of process buffers (14). In such case, few nonspecific identification tests—such as compendial methods, pH, osmolality, color/appearance used in combination with UV-visible spectrometry, high-performance liquid chromatography (HPLC), or other tests—may provide specific ID for similar media. This approach necessitates implementation of platform methods, which are applicable for several media with different composition. This approach ensures simplicity, cost effectiveness, and consistency between operators in various global companies. The approach also certifies quality of media in addition to tests provided on a vendor’s certificate of analysis (CoA) by continuously evaluating cell-culture media properties during release testing.
The goal of this work was to develop a platform approach and an algorithm for specific identification of seven culture media with essentially similar formulations. The study also includes development and validation of the platform method for quantitation of glucose as one of the key components of studied media.
Materials and methods
All eight multi-component culture media used in this study were custom manufactured for Bristol-Myers Squibb Co. The media samples were obtained as a powder and were reconstituted to the liquid stage. Seven media (M12, M51, M55, M10, M53, M57, and M17) had essentially similar formulations and were used in the development of identity testing strategy. The eighth media—media CDS, M12, and M57, with a target glucose concentration after reconstitution of 6.3 mg/mL, 2.0 mg/mL, and 3.0 mg/mL, respectively—were used in the glucose method validation.
The reagents used for glucose analysis were purchased from Sigma-Aldrich (now MilliporeSigma) in Saint Louis, MO: glucose 100 mg/mL (catalog number G8644), fructose (catalog number F0127), galactose (catalog number G6404), mannose (catalog number M2069), Sigma Protein-Free (SPF) media (catalog number C5467), and glucose hexokinase reagent (catalog number G3293). HPLC-grade water (catalog number AH-365-4) was obtained from Burdick and Jackson (Muskegon, MI).
The reagents for folic acid analysis were purchased from Fisher Scientific, Sigma-Aldrich, and other vendors and are not discussed in this work in detail. The United States Pharmacopeia (USP) reference standard of folic acid was purchased from Sigma-Aldrich (Saint Louis, MO) (catalog number 1286005).
Osmolality method
Osmolality was performed essentially following USP <785> (15).
Glucose method
Glucose standard curve—A stock solution of 100 mg/mL glucose was diluted with HPLC-grade water to produce a final concentration of glucose varying from 0.1 to 2.5 mg/mL. The contents of the hexokinase kit were reconstituted with 50 mL of HPLC-grade water and 10 μL of each of the diluted glucose sample was added to 1 mL of reconstituted hexokinase reagent. After incubation for 15 min, the absorbance at 340 nm (A340) was measured using an Agilent 8453 UV–VIS spectrophotometer. A plot of the A340 vs. glucose concentration (mg/mL) was analyzed by linear regression.
Analysis of culture media—The SPF culture media was analyzed as part of system suitability (described in the following passages). SPF culture media was diluted two-fold with HPLC-grade water and 10 μL of diluted media was analyzed in six replicates in a manner similar to a glucose standard. The average concentration and relative standard deviation (%RSD) were reported.
Analysis of test articles (culture media—The target concentration of glucose in the test article was used to calculate a dilution factor and each culture medium was diluted with HPLC-grade water to produce glucose concentration of ~2 mg/mL. Each test article was analyzed in triplicate.
System suitability requirements—The method has the following system suitability requirements:
• The coefficient of determination (R2) for the standard curve must be ≥ 0.98.
• The average glucose concentration in SPF media must be ±15 % of the value reported on the CoA supplied by the manufacturer.
• The %RSD of the glucose concentration from the six replicates of SPF media must be ≤ 15%.
The media sample analysis was initiated only if all system suitability requirements were met.
Folic acid method
The folic acid method utilized an ultra-high performance liquid chromatography (UHPLC) system from Waters Acquity equipped with photodiode array detector (PDA) and Acquity UHPLC column, BEH C18, 1.7 µm, 2.1 x 100 mm. A gradient separation was performed using mobile phase A: 50 mM phosphate, 4 mM 1-heptanesulfonic acid, pH 4.5 and mobile phase B: 30% acetonitrile and 10% methanol in mobile phase A (other method details are not included). The folic acid in the media was determined from a standard curve.
Results
Glucose method development and validation
As mentioned previously, the authors’ goal was to develop and validate a simple method for quantitation of glucose that first, uses common equipment available at various manufacturing sites worldwide and second, may serve as part of the specific identification platform for multi-component cell-culture media. For this purpose, a standard curve glucose method was developed and validated by UV-visible spectrometry, using similar principles as the commercially available kit. In the commercial kit, quantitative analysis of glucose is based on two sequential enzymatic reactions shown in the following passages (16, 17).
The first reaction is the phosphorylation of glucose to glucose-6-phosphate (G-6-P). This reaction is catalyzed by the enzyme hexokinase and utilizes adenosine triphosphate (ATP) as the source of phosphate. The G-6-P formed in the first reaction is oxidized to 6-phosphogluconate (6-PG) by the enzyme glucose-6-phosphate dehydrogenase (G-6-P DH). During this oxidation, an equimolar amount of nicotinamide adenine dinucleotide (NAD) is reduced to form nicotinamide adenine dinucleotide (NADH), which absorbs light at 340 nm and can be quantitatively assayed using UV-VIS spectrometry. The reaction scheme is shown in Equations 1 and 2.
Based on the stoichiometry of the reaction, the molar concentration of glucose is equivalent to the molar concentration of NADH. The concentration of NADH is determined by using Beer-Lambert’s Law.
A validation of the assay was performed in accordance with USP–NF General Chapter <1225>, International Council on Harmonization (ICH) Q2(R1) Tripartite Guideline Validation of Analytical Procedures: Text and Methodology, and FDA Guidance for Industry, Bioanalytical Method Validation (18, 19, 20). The parameters of the validation included specificity, linearity, accuracy, precision, range, limit of detection (LD), limit of quantitation (LQ), and robustness (21).
Specificity was demonstrated by analysis of three different multi-component media that contained glucose ranging from 2.0 to 6.4 mg/mL (Table I). In addition, the specificity of the method for glucose as compared with other hexoses was demonstrated by analysis of fructose, mannose, galactose analyzed at 2.5 mg/mL, and sucrose (a commonly used excipient) analyzed at 5 mg/mL. The results presented in Table I indicate that the experimentally determined glucose concentration in the three culture media (CDS, M12, and M57) are within 97—103% of the expected values, which meets the method acceptance criteria of 85–115% of the expected concentration. Among the hexoses and sucrose that were tested, only glucose was detected by this method, thereby demonstrating specificity for glucose (Table I). A second aspect of specificity was to demonstrate that media components do not interfere with the detection of glucose. This specificity can be accomplished by showing that there is no signal from media that contains all components except glucose. Because glucose-free media—which would work as a negative control for each studied media—is not commercially available; an alternate approach is described in the following passages.
Table I: Validation of the glucose method with specificity as a parameter.
Linearity of the signal was demonstrated from the standard curve (Table II and Figure 1A). The signal was linear as a function of glucose from 0.1 to 2.5 mg/mL in water. The coefficient of determination (R2) was 0.99, hence meeting the acceptance criteria of ≥ 0.98.
It was also necessary to confirm that the signal was linear as a function of glucose concentration in multi-component cell-culture media. The concentration of glucose in the cell-culture medium CDS is 6.4 mg/mL. Therefore, it was not feasible to spike in glucose at the levels that were studied in water. Therefore, an alternative approach was used to demonstrate linearity. As discussed previously, the method measures the absorbance of NADH, which is formed at an equimolar ratio with glucose (Equations 1 and 2). The linearity of signal produced at low levels of glucose in the media can, therefore, be assessed by spiking the media with NADH at concentrations that correspond to those typically generated by glucose at 0.1 to 2.5 mg/mL. The values of absorbance at 340 nm for NADH spiked in the CDS media are shown in Table II.
Table II: Validation of the glucose method with linearity as a parameter.
A graph of absorbance vs. concentration of NADH (mM) is shown in Figure 1B. The graph is linear with a coefficient of determination (R2) of 0.99. This experiment demonstrates the linearity of detection of NADH in the culture media.
Figure 1: Linearity of the glucose method. illustrates a glucose standard curve, and shows an overlay of a glucose standard curve with the curve for nicotinamide adenine dinucleotide and hydrogen (NADH) spiked in CDS cell-culture media.
An overlay of the absorbance of NADH spiked in media and the linearity curve of glucose in water (Figure 1B) indicates that linearity in media is comparable to that in water, as the corresponding slopes were linear (0.0575 vs. 0.0552 absorbance units/mM, respectively). The media components did not interfere with the detection of NADH, thereby demonstrating assay specificity. This approach confirmed that the assay was specific for glucose, and data generated for linearity with glucose in water can also be used to determine accuracy, range, and LQ of the method.
Accuracy was determined by calculating the percent recovery of glucose spiked in CDS culture medium at concentrations ranging from ~50% to ~150% of the glucose concentration in each medium. The spike recovery was 104–107% and within 85–115% of specified value. The precision (repeatability) was demonstrated by showing that the %RSD of six replicates of the measured concentration of glucose in CDS medium was 1%. Intermediate precision was measured by the analysis of three independent lots of CDS medium by two analysts over three different days. The %RSD for analyst 1 for days 1, 2, and 3 was 1% for each lot. The %RSD for analyst 2 for days 1, 2, and 3 was 3% for each lot. The overall RSD for three lots over three days within two analysts was 2%, which met the acceptance criteria of ≤ 15%. The LQ was experimentally shown to be 0.15 mg/mL. Based on the linearity, accuracy, and LQ, the range of the assay is 0.15 mg/mL to 2.5 mg/mL. The calculated LD was 0.07 mg/mL. Because the assay is based on the detection of NADH, which is generated in equimolar amounts to glucose, it was important to ensure that all the reactions go to completion. Therefore, a critical parameter is the time of incubation of the culture media with the hexokinase assay reagent kit. Robustness was demonstrated by showing that the results of three different media, in which the incubation time was varied from 15 min to 3 hours, were comparable (data not shown).
As mentioned previously, the glucose method was applied to M12 and M57 media (Table I) and the experimentally determined concentrations of glucose were within 15% of the expected values. An applicability of this method was also demonstrated for all other media used in the study (data not shown). Overall, the validated method is specific for a detection of glucose without matrix interference and may be used as a platform approach for multi-component cell culture media analysis.
Identification of culture media
For the development of an identification strategy, a few steps were taken to evaluate the methods and results for seven culture media, M12, M51, M55, M10, M53, M57, and M17, used in this study. All compendial microbiological tests performed for each media as part of release testing were excluded from this evaluation. In the first step, simple tests also provided in the vendor’s CoA, such as color and appearance, were evaluated. The results revealed that appearance of all media in powder form prior to reconstitution was essentially similar and varied from off-white orange-beige to pale orange color (data not shown). Therefore, this test cannot serve as a differentiator of media.
In the next step, FT–IR with attenuated total reflection (ATR) accessory was performed on all media in powder form as per USP <197> (4). The spectra did not show any significant difference and, therefore, cannot be used for specific identification of media (data not shown). The evaluation of solubility and pH values of media also did not show any significant difference and cannot be used for the discrimination of media (data not shown).
In the next step, the authors assessed osmolality values of seven media and developed respective specifications based on analysis of several lots of material (Table III). The osmolality values are also provided on vendor’s CoA. The specification range for the tests plays a crucial role in media identification, because it serves to differentiate the media containing the same components at different concentrations. A unique identification of each media is based on the principle that the specification range of each media is not overlapped with others. For the purpose of this study, the individual testing results are less important as long as their values fall within the specification range. A similar approach was employed to identify the buffers used in vaccine production (14).
Table III: Characteristics of cell-culture media by osmolality, glucose, and folic acid methods.
Based on the specification range presented in Table III, three media—M12, M51, and M55—were segregated from others, because their osmolality values were in the range of 125–165, 800–1050 and 366–496 mOsm/kg, respectively. They also were also distinguished from four other media (M10, M17, M53, and M57), which were identified as a group given that their overlapped osmolality specification values ranged from 225–348 mOsm/kg (Table III).
In the next step, the authors assessed target glucose values determined by the platform glucose method and developed respective specifications (Table III). The data and specifications were evaluated for all studied media including media individually discriminated by osmolality tests as well as media segregated as a group. Based on glucose specification ranges of 8.5–11.5, 2.5–3.5, and <0.05 mg/mL, three media—M53, M57, and M51, respectively—were individually segregated from others. Two media, M12 and M55 (identified from each other by osmolality test), had similar target content of glucose (2 mg/mL) and overlapping specification in the range of 1.7–2.3 mg/mL and 1.6–2.4 mg/mL, respectively, and were therefore segregated as a group. Two remaining media, M12 and M17, with overlapping glucose specification range of 3.8-6.4 mg/mL and 4.1–6.1 mg/mL, respectively (segregated as a group), were identified by the folic acid method. The specification ranges for folic acid for the media M10 and M17 were 3.2–4.8 mg/mL and 5.1–7.7 mg/mL, respectively.
Discussion
Aspects of cell-culture media identification
The authors’ approach for specific identification of cell-culture media as well as other complex non-compendial raw materials is based on evaluation of available tests provided by the vendor CoA, in-house methods, and established strategies for release testing, which comply with FDA, CFR 211.84, and other regulatory requirements (1, 2). Several aspects need to be considered in the development of the strategy. It is important to emphasize that simplicity related to the method performance in the QC environment, aspects of global technology transfer, and equivalency of the equipment used in testing need to be considered when this strategy is being developed. Also, the methods, which provide data that can be used without any further data processing, must be considered first. Prior to establishing a release testing scheme and specifications, all culture media properties with close formulations from one vendor or multiple vendors need to be evaluated. In addition, they need to be compared with the properties of other media used at the same manufacturing facility to ensure that all media are specifically identified. This approach would help support quality compliance of the facility. It is also important to pay attention to release tests of media from a single supplier, because it is easier to accidently substitute one media for another if they have similar formulations and labels and are from the same vendor.
Based on these aspects, the authors developed a general tool for cell-culture media identification, which include a combination of the following methods: osmolality (compendial), platform glucose by UV-visible spectroscopy, and complementary folic acid by UHPLC. The osmolality method established an alignment of all acquired data with that from the vendor’s CoA. Although, solubility, pH, and color appearance tests did not distinguish the media from one another by established specifications, all of these tests were used as part of release testing.
In light of the aforementioned approach, it was important to develop a simple platform test for quantitation of glucose, which is present in each one of the studied culture media, except one. Also, by employing a linear standard curve for a calculation of glucose concentration in media, the method is independent of the UV-visible spectrometer being used, which helps simplify method transfer. This method can also be adapted to a 96-well plate for high-throughput media screening. Validation of the assay for quantitation of glucose in the culture media includes the demonstration of specificity. This is normally performed by demonstrating the absence of a signal in the media that contains all components except the analyte of interest, which in this case is glucose. The lack of availability of glucose-free media necessitated an alternate approach to demonstrate specificity. As described in the introduction, the glucose levels are quantitated by measuring the NADH that is generated in presence of glucose (Equations 1 and 2). Thus, the absorbance of commercially purchased NADH spiked in water was shown to be comparable to the absorbance of NADH spiked in culture media, indicating that the media components do not interfere with the detection of glucose. The folic acid method was used as a supplementary technique for the two remaining cell-culture media that could not be distinguished by a combination of osmolality and glucose methods.
Strategy and algorithm for specific identification of culture media
Based on the acquired data and established specifications discussed in the previous section, a streamlined approach to identity testing was defined that will allow efficient, unambiguous identification of the seven cell-culture media used in this study. The strategy for specific identification of studied media is shown in Figure 2. The scheme demonstrates how data acquired from osmolality and glucose analysis, complemented with folic acid test when needed, results in unambiguous identification of all seven media. The osmolality test differentiated three studied media (M51, M12, and M55). The specifications of four other media (M10, M17, M53, and M57) overlapped and were between 225 mOsm/kg and 348 mOsm/kg. Therefore, these media were segregated together and identified as a “group identity” because their specifications overlapped (Table III). Glucose analysis of all seven media further discriminated two media (M53, M57) not segregated by osmolality and identified two “group identity” media, (M10, M17) and (M12, M55), which had overlapping specifications. The glucose specification and osmolality specification range of M51 media distinguished it from others. However, the M12 and M55 media discriminated from others by osmolality, belong to a “group identity” based on the similar glucose specification range (Table III). The remaining media—M10 and M17—are effectively identified by folic acid as a complementary test; only these two media were tested by folic acid. It is important to emphasize that all tests used in this strategy are non-specific and only a combination of two or more tests would provide a specific identity of each cell-culture media. Moreover, because the tests are non-specific, the established specification range for each test would serve as an actual discrimination tool in identification of media. Obviously, all data acquired by all methods need to fall within the respective specifications. Another important point is that due to this “non-specific” identity of each media by a single test, only a combination of two or more tests would provide a specific identity as required per CFR 211.84. The intermediate step of “group identity” is essential, because it narrows down a number of media for further identification.
Figure 2: Strategy for identity testing of the studied cell-culture media using osmolality, glucose, and folic acid methods. Non-specific identification of each media is defined by triangles.
The scheme presented in Figure 2 for studied media can be extrapolated to suggest a general algorithm for their identity testing. This more general scheme is shown in Figure 3. A similar approach was proposed for chromatography resins in another, separate study, but employed different methods (22). As mentioned previously, compendial microbiological tests were not included in this algorithm, but need to be performed as well as part of release testing. Glucose analysis can be pursued in parallel with the osmolality analysis to identify the media and confirm the quantity of glucose within established specifications. The quantitation of glucose is specifically important for media with similar composition manufactured and labeled by the same vendor. Any sample not identified through the combination of the described two non-specific identity tests can be subjected to folic acid analysis and quantity verification against established specifications to provide a final discriminative evaluation of sample identity.
Figure 3: General algoithm for identity testing of a collection of cell-culture media using a combination of non-specific methods including osmolality, glucose, and folic acid. In addition, label check, color and appearance, and pH and solubility tests are performed for each culture media.
Overall, the proposed testing scheme can incorporate a few relatively simple analytical methods into an efficient testing algorithm to provide definitive identity confirmation for cell-culture media from one or more vendors. The methods are well suited for routine use in a QC environment and the specific identification approach meets regulatory requirements. This is an example of a platform tool for glucose, which may be used for other strategies applied to similar media. It is important to highlight that glucose, as a more simple test compared with a folic acid test, may be used alone for the identification of five cell-culture media examined in the study, while only two media needed to be identified through the folic acid method.
Conclusion
The developed testing algorithm, which incorporates a combination of a few relatively simple analytical methods such as osmolality, quantitation of glucose, and folic acid, provides specific identity confirmation for seven studied cell-culture media with essentially similar composition. The methods are well-suited for routine use in a QC environment and the provided identification approach meets CFR, FDA, and other regulatory agency requirements. This assay used in this study was shown to be specific for detecting glucose in all studied multi-component media without matrix interference, regardless of a glucose concentration or the vendor.
Acknowledgements
The authors would like to thank Dr. Sam Mathew, Michael Adamo, and Jaimin Patel from Bristol-Myers Squibb, as well as Dr. K.C. Cheng (currently at Actinium Pharmaceuticals, Inc.) and Dr. Xiao-Ping Dai (currently at Celgene Corporation) for providing valuable input.
References
1. CFR Title 21, Part 211.84: Food and Drugs (Government Printing Office, Washington, DC), pp. 148–149.
2. G. Beck et al., BioProcess Int. 8 (7), pp. 2–13 (2009).
3. N. Afonina, B. Lay, Z. Akhunzada, S. Mathew, S. Mallya, and M. Grace, “Controlling Raw Materials in Biological Manufacture: Regulatory and Analytical Aspects,” presentation at Informa/IBC Life Sciences (Cologne, Germany, June 27–28, 2012).
4. USP, USP General Chapter <197>, “Spectrophotometric Identification Tests,” USP 38–NF 33 pp. 220–221.
5. N. Afonina, J. Timmermans, and L. Bhattacharyya, “A Novel Approach to the Development of ID Release Tests for Resins and Growth Media,” presentation at the 2nd Annual Meeting on Raw Materials and Contract Services for Mammalian Cell Products (St. Louis, MS, June, 11–13, 2000).
6. G. Reich, Adv. Drug Deliv. Rev. 57 (8), pp.1109–1143 (June 15, 2005).
7. T. Strother, European Biopharm. Rev. EBR 12 (169) article 3215, (2012).
8. H.W. Lee et al., Biotechnol. Prog. 28 (3), pp. 824–832 (2012).
9. A.O. Kirdar, G. Chen, and A.S. Rathore, Biotchnol. Prog. 26 (2), pp. 527–531 (2010)
10. C. Sharma et al., BMC Proceedings 5 Supplement (8), P5 (2011), doi:10.1186/1753-6561-5-S8-P5 (2009).
11. B.P.W. Ryan et al., Biotechnol. Bioeng. 107 (2), pp. 290–301 (2010).
12. J.K. Simonelli et al., Chromatographia 66 (11–12), pp. 977–981 (2007).
13. J. Zhang et al., Electrophoresis 30 (22), pp. 3971–3977 (2009).
14. N. Afonina, K. Bhatt, L. Howson, and L. Bhattacharyya, “ID Tests for Liquid Formulation Raw Materials: Growth Media and Buffers,” presentation at the 7th Annual Meeting on Viral Vectors and Vaccines (Lake Tahoe, NV, Nov. 6–9, 2000).
15. USP, USP General Chapter <785>, “Osmolality and Osmolarity,” USP 38–NF 33 (US Pharmacopeial Convention, Rockville, MD, Feb. 1, 2015) pp. 541–543.
16. J.J. Carroll, N. Smith, and A.J. Babson, Biochem. Med. 4, pp. 171–180 (1970).
17. M.W. Slein, Methods of Enzymatic Analysis, H.U. Bergmeyer, Ed. (Academic Press, New York, NY 1963), pp. 117–123.
18. USP, USP General Chapter <1225>, “Validation of Compendial Procedures,” USP 38–NF 33 (US Pharmacopeial Convention, Rockville, MD, Feb. 1, 2015) pp. 1445–1461.
19. ICH, Q2(R1) Tripartite Guideline Validation of Analytical Procedures: Text and Methodology, Step 4 version (2005).
20. FDA (CDER), Guidance for Industry, Bioanalytical Method Validation (Rockville, MD, May 2001).
21. A. Wierzbowski, L. McAleer, M. Jin, N. Afonina, and S. Mallya, “Validation of Analytical Methods,” presentation at the Stability Conference of International Pharmaceutical Association of Canada (New Brunswick, NJ, May 2013).
22. N. Afonina, L. Bhattacharyya, and J. P. Hennessey Jr., The Analyst (UK) 129, pp. 1091–1098 (2004).
About the Authors
Satish Mallya is senior research investigator; Benjamin Lay is lab supervisor; Lihong McAleer is scientist; Alexandria Emory is associate manager; all at Bristol-Myers Squibb. Nataliya Afonina is president and principal consultant at AN Biologics Consulting LLC.
ALL FIGURES ARE COURTESY OF THE AUTHORS.