Considerations When Implementing Automated Methods into GxP Laboratories
| A Tutorial from the Journal of the Association for Laboratory Automation|
Originally appearing in JALA 2005 10 182-191(view)
Considerations When Implementing Automated Methods into GxP Laboratories
Authored by: Gregory K. Webster, Laila Kott, Todd D. Maloney
In the increasingly scrutinized pharmaceutical industry, regulatory agencies are demanding validation of any and all analytical instrumentation, including documentation associated with its implementation, qualification, and ability to report accurate and reliable results. Herein, we discuss the qualification and validation of an automated liquid handling system and an automated dissolution method. We describe the comparison of automated experiments versus manual experiments while addressing the pertinent validation and qualification considerations for each. Discussion of documentation and validation required for various regulated laboratories (good clinical practices (GCP), good laboratory practices (GLP), and good manufacturing practices (GMP)) is also reviewed.
It is not a new aspect of the industry; competition heats up and companies respond in one of two ways. By reducing resources, they strive to maintain their cost effectiveness and profitability to get the job done. This can occur through layoffs, attrition, and budget freezes. Companies reduce expenses and employees are asked to get good at “doing more with less.” Alternatively, companies can “work smarter not harder.” In some circumstances automation can provide strategies for improving cycle times and dramatically increasing the capacity of laboratories. However, automation is not a panacea and requires a careful analysis of the total sample stream workflow.
Modern analytical laboratories are full of significant technological investment. Many are surprised to learn that the highest cost in a laboratory is still not the equipment but the analyst, and more specifically, analyst time. With analysts spending as much as 60% of their time on sample preparation, it is no wonder companies are turning to automation to recapture their investment in human resources. Only through a careful analysis of what tests, or portions of tests, are automated will one be able to assess the return on investment from automation. In the worst cases, an analyst could be replaced with an equipment repair technician, or a step that is not a critical path in the laboratory could be automated. In the best case, significant additional testing capacity and reduced cycle times can be created.
Both routine and intensive analyst tasks are being automated because of favorable financial return. However, for laboratories in regulated industries, using this automation is not as effortless as simply turning on the switch. A significant amount of planning, along with additional investment in validation and documentation, must be invested prior to relying upon automation to generate a return.
In the United States, laboratories testing products regulated by the Food and Drug Administration (FDA) or the Environmental Protection Agency (EPA) must operate within the guidelines published by these agencies and codified as law in the Code of Federal Regulations (CFR) and Federal Register (FR). These guidelines are known as Good Manufacturing Practice (GMP), Good Laboratory Practice (GLP), and Good Clinical Practice (GCP). GMP guidelines are codified under 21 CFR Parts 210 and 211 for finished pharmaceuticals, and 21 CFR Parts 225 and 226 for Type A medicated articles and finished feeds regulated by FDA. GLP guidelines are codified under 21 CFR Part 58 for nonclinical laboratory studies under FDA. The EPA regulates GLP through 40 CFR Part 160 for products/laboratory analysis regulated by the Federal Insecticide, Fungicide, and Rodenticide Act of 1996 (FIFRA) and 40 CFR Part 792 160 for products/laboratory analysis regulated by the Toxic Substances Control Act of 1976 (TSCA). Global acceptance for GCPs written by the International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH) have been published, and thus are codified under 62 FR 25692 for products/laboratory analysis regulated by FDA. For clinical laboratories, GCPs are codified under 21 CFR Part 493, the Clinical Laboratories Improvement Act of 2003 (CLIA). GCPs can often be interpreted for the needs of each business concern. If work is being done in these areas, it is recommended that the quality assurance department be consulted to determine which GxP regulations apply.
GMP, GLP, and GCP interpretations for analytical laboratories contain many common elements and have now become collectively known in the pharmaceutical industry as “GxP”. Grouping these guidelines as GxP compliance has created a harmonized norm and helped many companies address the varied interpretations that exist within the industry. In terms of instrumentation, it is arguable whether GMP or GLP is stricter in their requirements, simply because their interpretation is left up to the business concern and each company interprets the guidelines to suit their own business practices. Setting up policies to comply with both GMP and GLP as GxP guidelines eliminates this issue within many companies.
Whether automated or not, GxP guidelines require analytical methods to be scientifically valid prior to implementation. The fact that automation can run more precisely or more accurately than manual methods is irrelevant. Documentation must exist for regulatory inspection that clearly demonstrates that the automated or manual method is appropriate for the phase of development and is sufficiently accurate, precise, and robust. Even though the compendial methods published by United States Pharmacopoeia (USP), American Society for Testing and Materials (ASTM), and Association of Analytical Communities (AOAC) have been validated and fully tested prior to publication, it is still up to the user to demonstrate and document that the method is performing properly on their equipment.
Details for a successful instrument qualification, operational qualification, and performance qualification, known collectively in the pharmaceutical industry as IQ/OQ/PQ, must be established under GxP. Table 1 lists the general definitions of the qualifications, with the added definitions for equipment qualification (EQ) and design qualification (DQ). (EQ and DQ are newer interpretations of equipment qualification and not as widespread in application.) Why all the fuss about instrument qualification? GxPs mandate that analysts establish and document procedures to indicate that the instrumentation has been fully controlled. The industry adage “if the scientist didn't write it down, you didn't do it” applies not only to data but to instrumentation documentation as well. For now, let it suffice to say that all measurement equipment in a GxP laboratory must be qualified. Automated equipment not only must be qualified to demonstrate its utility, but also to address concerns from directing much of an analysis away from human intervention and control.
|Table 1. Qualification definitions|
|Equipment Qualification (EQ)||EQ is the process of ensuring that an instrument is appropriate for its intended use and that it performs according to the specifications agreed upon by the user and supplier.|
|Design Qualification (DQ)||DQ defines the functional and operational specifications of the instrument and details the conscious decisions in the selection of the supplier prior to the installation of the system|
|Installation Qualification (IQ)||IQ establishes that the instrument is received as designed and specified. IQ also verifies that the instrument environment is suitable for the operation and use of the instrument and that the instrument is properly installed.|
|Operational Qualification (OQ)||OQ is the demonstration that an instrument will function according to its specifications in the selected environment.|
|Performance Qualification (PQ)||PQ is the demonstration that the instrument performs according to a specification appropriate for its routine use.|
Validation vs. qualification
Validation is the process of evaluating the performance of a specific measuring procedure and checking that the performance meets certain preset criteria. Validation establishes and provides documented evidence that the measuring procedure is fit for a particular purpose. Qualification is the verification that an instrument is performing under predetermined specifications. It should also be noted that “calibration” is also unique; it is the verification that an instrument is standardized against a nationally accepted standard (or a secondary standard that has been calibrated against such a standard). The terms “validation” and “qualification” are often mistakenly interchanged.
Regulated industries look for equipment to undergo a series of qualifications: Equipment Qualification (EQ), Design Qualification (DQ), Installation Qualification (IQ), Performance Qualification (PQ), and Operational Qualification (OQ). These qualifications are defined in Table 1. Specifications for these qualifications are often predefined by the manufacturer and negotiated within the company.
All methods used for GxP investigations are required to undergo stage-appropriate validation to demonstrate confidence in the analytical result prior to reporting. In regulated industries, this confidence only exists after the method has been evaluated to industry guidelines and the investigation documented appropriately.
Automated methods can be introduced to the GxP laboratory by (1) validating the automated method directly or (2) establishing a successful analytical method transfer from an existing method to the automated procedure.
The level to which a method is validated for GxP is normally based on where the methodology is being applied. In early drug development, there is less information about the chemical characteristics of the drug and a high likelihood it will not succeed and become a marketed product. It is not economically prudent to allocate significant resources to the drug, and the level of method validation is to a level consistent with (a) the state of knowledge regarding the manufacturing processes and (2) the concern for safety of patients taking the drug in a clinical setting. Reference standards are seldom available at this stage, leaving the validation of the method focuses on precision and selectivity of the drug from its matrix. Linearity of response is needed for all stages of development; however, sensitivity only needs to be established to the reporting limit. If the GxP study involves human testing, the level of validation increases to a point where confidence in the designed indication going into the patient is established. Finally, as the drug progresses to market, full analytical characterization is established.
Analytical Validation Principles
Validation is establishing documented evidence that the analytical test method performs to established specifications. These specifications are generally defined in a study protocol, in company standard operating procedure (SOP), or in a compendial reference. If the method is being used for stability investigations, the validation must establish its suitability as a stability-indicating method for the analysis of the intended analyte.
To some degree, all analytical method validations under GxP will test for specificity, precision (repeatability and intermediate precision), linearity, range, accuracy, robustness, and quantitation limit of the intended analyte.
Check for interferences: Placebo samples for the method are prepared and spiked individually with the intended analyte (and degradation products, if available) to check for interferences. No interfering response should be detected.
Accelerated degradation/peak purity: For separation methods, accelerated degradation studies are run using the analyte and intended matrix. Samples are generally degraded by exposure to acidic (pH 1–6), basic (pH 8–13), oxidative, light, elevated temperature (90°C), and humidity conditions. The conditions are adjusted, as necessary, in order to attempt degradation of the analyte until there is an approximate 20% loss relative to the analyte in the initial sample. No interfering response should be detected.
Resolution factor: A minimum resolution factor will be determined for the degradation studies. Not all peaks detected may end up to be baseline resolved; however, the analyte of interest must remain baseline resolved. The resolution factor will be established based on the least acceptable overall minimum separation.
Linearity and range
Linearity is established from a minimum of five analytical standards from 50–200% of the nominal concentration of the analyte standard. The exact range and concentration levels are generally established by company policy. For impurity determinations, a second analytical curve is often generated at 0.1% range of the nominal API standard concentration.
For devices known not to respond in a linear fashion, additional studies need to be incorporated to demonstrate the method responses to the analyst of interest in a proportional manner. As an example, several chemical sensors respond in a nonlinear fashion. Their use is often incorporated with polynomial regressions and suitability established with thorough accuracy investigations.
The accuracy of the method with respect to the quantitation of the intended analyte is established in any of three procedures:
Assay of previously qualified samples: In this mode, the accuracy of the analytical method is established by comparing the results of the method being validated against the certified standard. Acceptable limits for the recovery to the certified value are generally established prior to execution of this study.
Assay of formulated samples: In this mode, the accuracy of the analytical method is established by analyzing samples formulated in an 80–120% range in the study matrix. The results of the method being validated are compared to the formulation concentrations. Again, acceptable limits for the recovery to the certified value are generally established prior to execution of this study.
In the absence of suitable standard materials: The accuracy of the method in early stages of product development is inferred if the criteria for precision, linearity, and specificity with respect to intended analyte is met.
The standard deviation, relative standard deviation, and confidence interval obtained for each precision determination is established.
Repeatability: Six replicate samples (masses, volumes) will be prepared from a single sample of the analyte in the intended matrix. The repeatability (method precision) will be determined using the test method. The relative standard deviation (RSD) specification for this investigation is often set at ≤2.0%, based on the type of analytical technique being employed.
System precision: Six replicate trials (injections) of a single sample are used to determine system precision. The RSD specification for this investigation often set at ≤−2.0%, based on the type of analytical technique being employed.
Intermediate precision/ruggedness: A second analytical system will repeat the accuracy and precision portions of the validation using a different instrument and different lots of reagents, as possible, to determine the intermediate precision (intralaboratory check). The results obtained by the two analysts for the accuracy determination should agree within 2%, based on the type of analytical technique being employed.
For analytical method that issues a quantitative result, the range or lower limit of quantitation of the method must be established. One simple procedure for doing this is as follows:
- Prepare two separate calibration curves by using duplicate standards, prepared at five different levels, in the estimated range of the quantitation limit.
- Generate the two calibration curves using least squares linear regression analysis. For each curve, determine the slope and the residual standard deviation of the regression line.
- Calculate the quantitation limit (QL) for each curve using the following equation:
Where σ = the residual standard deviation of the regression line; m = the slope of the regression line
- Calculate the average of the QL values obtained from the two calibration curves. Confirm that the calculated quantitation limit is a concentration that is accurately quantifiable by preparing three solutions at that level and analyzing them against a standard.
- The experimental value should be within 10% of the average prepared value. If this specification is not met, the quantitation limit will be increased by 50% until a concentration is reached that meets this criterion.
- Confirmation of this quantitation limit will be established by the analysis of three additional drug sample preparations at the established quantitation limit level.
The robustness of an analytical method can be established by verifying that the method still acts in a suitable manner in the presence of small changes in method conditions. For the validation, method conditions are altered one condition at a time. Possible changes:
- ±2% relative change in the volume of the lesser component (organic or aqueous) of the mobile phase (the larger component volume remains unchanged)
- ±2 °C change in system temperature
- ±5% relative change in the system flow rates
- ±0.1 unit in solvent pH
Many automated methods exist simply to make execution of a manual method more efficient. In this regard, the use of an automated method should be seen as an analytical method transfer from the established method. In GxP, the transfer of an analytical method must demonstrate that the results obtained by a secondary method or location are not significantly different from the results obtained using the method running under its established validated conditions.13 Accordingly, USP in their “Automated Methods of Analysis” General Chapter 16, agrees that a method does not need to be revalidated after it is automated, but that the notion of an effective method transfer still applies for compendial methods:4
Before an automated method for testing an article is adopted as an alternative, it is advisable to ascertain that the results obtained by the automated method are equivalent in accuracy and precision to those obtained by the prescribed Pharmacopeial method, bearing in mind the further principle stated in the General Notices and Requirements that “where a difference appears, or in the event of dispute, only the result obtained by the procedure given in this Pharmacopeia is conclusive.
For transferring a method to an automated system, the original method and automated procedure laboratories evaluate the methods and agree upon the transfer process in accordance with company procedures. The data generated by the automated procedure is compared to that generated by the original procedure, using predetermined testing pattern and acceptance criteria. The original method validation data should be reviewed to determine the applicability of the acceptance criteria. The need for training prior to generation of comparative data is determined on a case-by-case basis and may consist of verbal instructions, demonstrations, and/or hands-on performance.
The analytical method transfer establishes documented evidence that the analytical test method performs in the same manner as the original manual method. The transfer specifications are generally defined in a study protocol or company standard operating procedure (SOP). Both the original manual method and the automated method must first be validated individually. The automated method should be validated under the same protocol as the original method. However, simply validating the automated method is not sufficient; the method transfer confirms that the results from the original and automated method are not significantly different.
The typical method transfer protocol entails the analysis of several samples of the intended analyte and matrix initially characterized by the original method and analyzed by the automated method. The automated method must establish and maintain the same suitability as the original method. The results are then statistically compared for equivalence. Failure to establish equivalence results in additional qualification of the automated method and a repeat of the transfer investigation.
System vs. component validation
In regulated industries, the question often arises whether the automated instrument should be qualified as a whole performing system (sometimes referred to as “holistic testing”) or qualified by module-by-module inspection (sometimes referred to as “modular testing”). The recommendation by FDA and Laboratory of the Government Chemist in EURACHEM (LGC/EURACHEM) is to use, in general, the holistic approach.  That is, the system is qualified as a whole and individual modules are investigated only if the system does not pass.
Case study I: Qualification of an automated liquid handling system
Prior to implanting a liquid handling system into several analytical procedures, qualification studies were performed to verify the fit for purpose of an automated liquid handling system. This investigation qualified that the instrumentation is suitable for use with additional GxP sample handling procedures. Additional studies will still be needed to authenticate the suitability of each individual procedure, as with any GxP method. However, the individual components and liquid transfer on the deck will no longer need to be qualified as long as the deck is maintained in its qualified state and no significant changes made.
The probe analyte for this investigation was caffeine. The effectiveness of sample transfer using this probe analyte was verified by an isocratic high performance liquid chromatography (HPLC) analysis with the conditions listed in Table 2.
|Table 2. Chromatographic conditions for the analysis of caffeine samples prepared by liquid handling system|
|Mobile phase A||0.2% perchloric acid|
|Mobile phase B||Acetonitrile|
|Column||YMC Pack Pro 3 um C18, 150 x 4.6 mm|
|Flow rate||1 mL/min|
The liquid handling system was configured to deliver standard aliquots of caffeine and diluent to prepare exact analytical standard preparations. Verification of the accuracy and precision of the automated liquid handle was verified by analyzing these test preparations by HPLC. Prior to analysis the chromatographic system was verified to yield a system precision of not more than 1.0% RSD.
Qualification of the Automated Liquid Handler
The liquid handler was programmed to prepare exact dilutions in the range of 5% to 150% of the nominal assay concentration. (0.0125–0.500 mg/mL) Each preparation was tested by HPLC. The accuracy of the liquid handler was verification with a linear regression analysis of the sample preps, illustrated in Figure 1. The resulting correlation coefficient for the actual versus theoretical response was r=0.9995, verifying excellent accuracy for the system.
|Figure 1. Linearity of response for caffeine samples prepared by the liquid handling system.|
The liquid handler was programmed to prepare exact dilutions in the range of 50% to 150% of the nominal assay concentration. (0.125–0.375 mg/mL) Each preparation was again tested by HPLC. The precision of the liquid handler was verified as charted in Table 3. The low relative standard deviations posted confirm the excellent precision achieved with the automated system.
|Table 3. Caffeine precision analysis|
|Nominal Standard Preparation||Concentration (mg/mL)||%RSD|
|RSD=relative standard deviation.|
A major concern with automated systems is their ability to repeat a procedure consistently. This concern for the liquid handler was alleviated by preparing 32 replicate sample preparations at 0.25 mg/mL and testing each preparation by HPLC. The resulting caffeine response yielded a 0.47% RSD for peak areas. The data verified the system is quite robust.
Case Study I: Conclusion
Based on the results presented in this section, the automatic liquid handler system is accurate, precise, and rugged when evaluated by liquid chromatography. The system is qualified for use with GxP determinations after incorporation into a suitably validated procedure.
Case study II: Validation of automated dissolution method by comparison to manual sampling
After a manual dissolution method was validated for a commercial dosage form, the method was transferred to an automated system. For ease of comparison, all method parameters were kept the same except for the sampling system. Only those elements that could be affected by the sampling system (selectivity, system suitability, precision, accuracy/recovery, and ruggedness) were studied. All samples were analyzed by HPLC with a UV detector using a certified reference standard for quantitation.
System suitability was tested with USP standard tablets: prednisone (Lot N) and salicylic acid tablets (Lot O). The lowest strength and highest strength of the commercial solid dosage form were used to bracket all strengths for evaluation of method precision and intermediate precision. Bulk drug and the excipient blend were used for accuracy/recovery.
Dissolution was monitored by dropping the tablets into a dissolution media and sampling every 15 min for 1 h. Dissolution profiles were collected in two separate experiments at paddle speeds of 50 and 75 rpm. The prednisone and salicylic acid samples were analyzed directly by UV and a commercial solid dosage form (CSDF) was analyzed by HPLC with a UV detector. Manual sampling was done with plastic syringes and metal cannulas. Samples were filtered post sampling directly into HPLC vials. Automatic sampling collected the same volume as was done manually and filtered the solutions on-line before being collected in test tubes.
Comparing Manual and Automatic Sampling
An f2 comparison was used to evaluate the similarity between the dissolution profiles obtained from manual and automatic sampling for both dosage strengths at 50 and 75 rpm. The fit factor, shown below, is a logarithmic transformation of the sum of squares error. The sum of squares between a reference profile (which, in this comparison, is the automatic sampling profile) and a test profile (manual pulls) are averaged and scaled to be between 0 and 100.
When the fit factor (f2) is 100, the profiles are identical. When the fit factor is 50, the profiles differ by 10% and the similarities continue to decrease as f2 approaches zero. The criterion for most comparisons and the current comparison is a 10% difference in profiles (a fit factor of 50).
Excipient mixtures (all components in the same proportion as in the final tablet) were prepared and evaluated as per the dissolution method. Three sample solutions were obtained by automatic sampling aliquots at 45 min and were filtered directly into test tubes, with no discard volume. No interfering peaks were observed at the retention time for the drug in the chromatograms. This result also demonstrated the lack of interference from the syringes and syringe filters used in automatic sampler.
Standard system suitability tests were performed using USP disintegrating calibrator tablets (prednisone) and USP nondisintegrating calibrator tablets (salicylic acid). The criteria for Lot N of prednisone is 28–54% dissolved at 30 min and for Lot O of salicylic acid is 17–26% dissolved at 30 min. The criteria for both the prednisone and the salicylic acid tablets were met, using both the manual and the automatic sampling. Statistical analysis showed that the data was normally distributed (Fig. 2). The variances for manual and automatic pulls were the same, and the means between manual and automatic sampling were not significantly different. Table 4 shows the data for each vessel: the averages, percent dissolved, and the specification.
| Figure 2. Distribution of system suitability data for both automatic and manual sampling for both (A) prednisone and (B) salicylic acid.|
|Table 4. System suitability results for prednisone and salicylic acid, with manual vs. automatic sampling|
|Prednisone (absorbance)||Salicylic acid (absorbance)|
|Specification:||28–54% dissolved||17–26% dissolved|
The system precision was evaluated by making six replicate automatic sample pulls at 75 rpm and 75 min. This sampling time point, which was longer than the typical 60 minute final pull, ensured complete drug dissolution and enabled evaluation of method precision independent of tablet dissolution variability. The samples were analyzed by HPLC and the relative standard deviations (%RSD) of the peak-area responses were calculated. The %RSD of the six automatic sample pulls based on percent label claim for both dosage strengths was 1.3 and 0.4 for the low and high dose, respectively. These met the predetermined criterion of %RSD ≤4.0.
The pure drug in the solid state was spiked into excipient mixtures simulating the test preparations of the lowest and highest strength tablets. For both strengths, three excipient samples were spiked with pure drug substance at about 50% of label claim, six at 100% label claim, and six at 125% label claim. These spiked samples bracket the expected concentration range of dissolution samples. These mixtures were added to the dissolution media and samples were pulled at 45 min. The results from the recovery samples are shown in Table 5. The procedure met the protocol acceptance criteria that the mean value obtained for each set of preparations is within ±5.0% absolute of the theoretical value and the %RSD of six spiked samples at 100% label claim for both dosage strengths is ≤4.0%.
|Table 5. Bulk drug recovery from spiked excipient blends|
|Percent recovery from blends spiked at|
|Lowest strength (LC) Highest strength (LC)|
Dissolution samples of six tablets from the same lots of the low and high dosage strengths were analyzed by two different analysts on different days using different dissolution baths and different automatic sampling systems. Table 6 shows the results for percent label claim dissolved at 45 min for the two analysts at 75 rpm paddle speed. These data met the acceptance criteria for agreement of mean results for each analyst. The data obtained at 45 min agree for both samples, within ±5.0% on an absolute basis and the %RSD of the second analyst is ≤4.0%.
|Table 6. Intermediate precision for percent dissolved at 45 min for the low and high strengths of the CSDF at 75 rpm|
|Percent label claim dissolved|
|Low Dose||High Dose|
|% RSD Analyst 2||3.0||0.4|
Figure 3, Figure 4 show the overlaid profiles of the manual and automatic sampling for both dosage strengths and paddle speeds. The comparisons all resulted in high f2 values ranging from 85.3 to 98.6 (see Table 7) and illustrate that manual and automatic sampling are comparable.
| Figure 3. Dissolution profiles of the lowest strength of the CSDF for both automatic and manual sampling at 50 rpm and 75 rpm.|
|Figure 4. Dissolution profiles of the highest strength of the CSDF for both automatic and manual sampling at 50 rpm and 75 rpm.|
|Table 7. f2 Comparison values for the low and high strengths of the CSDF at 50 and 75 rpm|
|Dosage strength||50 rpm||75 rpm|
|Profiles are not significantly different if f ≥ 50|
Case Study II: Conclusion
Based on the results presented in this report, the automatic sampling systems are equivalent to manual sampling when evaluating dissolution profiles of the commercial solid dosage form's, and therefore can be used in tandem or instead of manual sampling for dissolution testing. In this manner the automatic sampling system method is also fully transferred and validated.
As demand for output from laboratories increases and budgets grow tighter, companies and their laboratories are automating more functions. The issues addressed in this paper revolve around the transfer of manually run methods to automated systems, especially in regulated environments. As discussed and shown in the case studies, the systems must be properly qualified, and the analytical methods used with the automated system need to be validated. Since many automated methods are employed simply to reduce analyst time and make manual methods more efficient, one can consider the validation of the automated analytical method as a simple method transfer exercise. This is supported by the USP in General Chapter 16, “Automated Methods of Analysis”, which states that a method needs not to be revalidated after it is automated, but a successful method transfer yielding results similar to those established by the non-automated method is required. Method transfer exercises are often governed by internal standard operating procedures, but typically entail showing the same suitability as the original method, analyzing several samples by both methods, and statistically comparing them to prove equivalence.
The authors would like to thank Zhaohui Lei and Teresa Lints of Pfizer for experimental contributions and Mark VanArendonk of Pfizer for technical background and discussions.
Analytical Research & Development, Michigan Pharmaceutical Sciences, Pfizer Global Research and Development, 2800 Plymouth Road, Ann Arbor, MI
Correspondence: Dr. Gregory K. Webster, Analytical Research & Development, Michigan Pharmaceutical Sciences, Pfizer Global Research and Development, 2800 Plymouth Road, Ann Arbor, MI 48105; Phone: +1.734.622.3848
- ↑ Guideline on the Validation of Analytical Procedures: Methodology (1). International Conference on Harmonization of Technical Requirements for the Registration of Drugs for Human Use, Geneva, Switzerland, May 9, 1997.
- ↑ VICH GL2. International Cooperation on Harmonization of Technical Requirements for Registration of Veterinary Medicinal Products, Brussels, Belgium, October 1998.
- ↑ Code of Federal Regulations, Title 21, Foods and Drugs (U.S. Government Printing Office, Washington D.C., 1 April 1997), Part 211.
- ↑ USP 25 (United States Pharmacopeial Convention, Rockville, Maryland, 2005).
- ↑ Green JM. Anal. Chem. 1996;68:305A–309A.
- ↑ Swartz ME, Krull IS. Pharm. Technol. 1998;22(3):104–119.
- ↑ Weiser WE. Analytical validation. Pharm. Technol. 1998;20–29.
- ↑ Guidance for Industry: Analytical Procedures and Methods Validation Chemistry, Manufacturing and Controls Documentation. U.S. Food and Drug Administration, U.S. Government Printing Office: Washington, DC, 2000.
- ↑ 9.0 9.1 9.2 Bedson P, Sargent M. Accred. Qual. Assur. 1996;1:265–274.
- ↑ 10.0 10.1 Bansal SK, Layloff T, Bush ED, Hamilton M, Hankinson EA, Landy JS, et al. AAPS Pharm. Sci. Tech. 2004;5(1):Article 22.
- ↑ Green JM. Anal. Chem. 1996;68:305A–309A.
- ↑ Webster GK, Farrand DA, Litchman MA. J. of Process Analytical Chemistry. 2003;8(1):11–17.
- ↑ Furman WB, Layloff T, Tetzlaff RF. J. AOAC Int. 1994;77:1314–1318.
- ↑ Bolton S. Pharmaceutical Statistics: Practical & Clinical Applications. 3rd ed.. New York: Marcel Dekker; 1997;.
- ↑ Seo PR, Shah VP, Polli JE. Pharm. Dev. Tech. 2002;7:257–265.
- ↑ Moore JW, Flanner HH. Pharm. Technol. 1996;20(6):64–74.
|Click [+] for other articles on||The Market Place for Lab Automation & Screening||The Market Place|
|Click [+] for other articles on||The Market Place for Lab Automation & Screening||The Market Place|