Analytical Quality by Design in the Pharmaceutical Industry

As Quality by Design has become a more established approach in pharmaceutical development and manufacturing, more companies are now integrating the concept into analytical method development. LCGC spoke to Rosario LoBrutto of Teva about the business of analytical QbD, the steps involved in implementing it, and advice to those who want to get started.

What is the goal of applying Analytical Quality by Design (AQbD) to analytical method development?

The goal of Analytical Quality by Design (AQbD) is to design a method that consistently delivers predefined objectives and to control the quality attributes of the drug substance and drug product. Implementation of QbD for analytical methods during development allows for enhanced understanding of the analytical method focusing on robustness and ruggedness designed with the end user in mind, thereby facilitating methods transfer and providing opportunities for continual improvement.

Why is Analytical QbD (AQbD) necessary or helpful? Isn't the traditional approach to analytical method development effective?

Analytical Quality by Design (AQbD) implementation is aligned with a risk- and science-based approach towards methods development and methods validation. The type and extent of the risk analysis depends on the stage of the project in the development timeline. Also, it requires the right mindset, using the right tools and performing the appropriate amount of work at the proper time during the development timeline. It also includes being proactive based on risk assessment outcomes.

The outcome of the systematic risk analysis allows for the identification of the potential critical analytical method variables that could be evaluated and refined using a statistical design of experiments. Statistical design provides an economical and efficient use of resources, when many method variables exist and provides a greater opportunity of finding optimum conditions from a large amount of data generated from a limited number of experiments. Statistical design and data analysis facilitates an in-depth understanding of the method and justifies the choice of ranges for variables and finds a robust (optimum) region for final method based on model generated.

What are the foreseen business benefits of applying Analytical QbD?

Applying QbD allows for reduced risk of method failures during release or stability testing and aids “out of specification” investigations. Ultimately, it increases overall quality and reduces costs.

By integrating AQbD elements into a development strategy, the critical sources of analytical variability (continuous and discontinuous) are identified through the appropriate risk assessments, then measured and understood so that they can be controlled with the appropriate control strategy. The resulting business benefits could include: (a) A structured approach aligned with a science- and risk-based approach towards method development and methods validation, especially for robustness and ruggedness; (b) An appropriate control strategy for critical method variables, which will enhance method performance and robustness; (c) A reduction in system suitability failures; (d) Lowered operating costs from fewer failures and deviation investigations; and (e) Faster technical transfer of methods between development and manufacturing site(s).

These benefits could translate into significant reductions in working capital requirements, resource costs and non-value added time.

What are the key stages in implementing a QbD approach to analytical method development?

First identify the quality analytical profile (QAP) that includes the list and reasoning for all required methods and initial specifications to manufacture and control the intended drug product or substance. In the QAP a review of the available information is included from external and internal sources. Then for each analytical method the quality target method profile (QTMP) is developed, which includes all method specific related information. This QTMP is updated as the method basis is enlarged and the knowledge accumulated.

A generic Ishikawa diagram (fishbone diagram or “cause-and-effect” diagram) for a particular analytical technique is then reviewed for further evaluation. The fishbone lists the inputs (causes) such as the method variables and the output (effects) such as the critical method attributes. The critical method attributes (CMA) are then defined based on the most recent version of the QTMP. A critical method attribute is an element of method performance that must be measured to assess whether a method is capable of producing fit-for-purpose data. One method variable may impact multiple CMAs. Multiple method variables may impact a particular CMA.

The project specific fishbone diagram is then constructed, mainly updating the relevant inputs (method variables) that are pertinent to the particular analytical technique. Selected method variables are then subjected to a risk assessment. A traffic light risk analysis matrix table is used for this initial risk assessment (using red, yellow and green colours to denote the level of anticipated risk, with red as the highest and green as the lowest). Method variables are defined as critical method variables (CMV) if they are deemed to have an impact on the critical method attribute (that is, identified as yellow or red in the traffic light risk analysis) as a result of the potential variable change from target. Once the potential critical analytical method variables are defined, then a screening design of experiment (DOE) can be performed to confirm and refine critical method variables based on statistical significance.

DOE is used to provide the most efficient and statistically sound approach to evaluate multiple method variables, their interactions and their responses (critical method attributes). This provides an excellent opportunity for screening a number of conditions generated from a limited number of experiments used in the first DOE study. Then as part of the DOE 1 data evaluation, statistical tools are used to identify critical method variables and the appropriate choice of ranges for method variables where a robust region for the critical method attributes could be obtained.

Then once the process is finalized, a failure modes and effects analysis (FMEA) approach can be used prior to final methods validation (for analytical methods to be used for release of Phase 3 clinical supplies of the brand product; and for analytical methods to be used for release of batch(es) used for supportive stability studies for ANDA submission of generic products).

A failure modes and effects analysis (FMEA) is a risk analysis tool that can be used for identifying the critical method variables, and the impact of analytical method variables on the critical material attributes identified in the fishbone.

The FMEA generally evaluates the potential deviations or failure modes of the variable from its set point. Each of the variables and material attributes are critically evaluated for their probability, (P), (probability that the variable would deviate from the target based on past experience or method understanding), detectability, (D), (modes of detection in current state) and severity, (S), (if it would occur what kind of effect would it have on the method performance or impact on the patient). For each (P), (S) and (D) a numbering system of 1-X is used to rank them accordingly, with X having the highest risk. For example, a higher number for detectability indicates that the deviation or failure is not easily detected, or the modes of detection are few.

The risk priority number (RPN) is calculated as the contribution of the three factors, the probability (P), the severity (S) and the detectability (D) of the variable evaluated as critical (CMV- critical method variable). The calculation formula is RPN = P x S x D. The number describes the level of risk associated with a specific failure mode. A lower RPN number implies less risk. If the RPN number is greater than a certain predefined threshold then actions must be taken to reduce the RPN to less than the threshold value. If a method variable has not been studied and the impact cannot be derived from prior knowledge or method experience this also should be given the highest priority for corrective action. The goal of these actions is to reduce the RPN score below the threshold level by identifying actions leading to improvement.

A method optimization DOE 2 could then be performed and statistical tools are used to identify the optimized choice of ranges for method variables based on model generated where a robust region for the critical method attributes can be obtained. A control strategy is then put in place for critical method variables in the analytical methods, and the FMEA updated and the methods validation completed. Finally an analytical method design space can be developed encompassing the proven acceptable ranges for the potential critical analytical method variables and their interactions and suitable control strategies can be implemented.

As the method continues to evolve and is used in different environments, continual improvement can be applied. The formulation and manufacturing process improvement should be accompanied by the respective analytical improvement as an outcome of the method validation process and additional critical attributes identified and additional data accumulated. The method control strategy evaluation should determine if the method serves its intended use, if the overall understanding of the method can improve its performance, or if, alternatively, changes should be made to the method or if it should be replaced by one more fit-to-purpose.

What advice would you give to someone who is just starting to apply QbD to analytical method development?

Adopt a mindset that moves away from reactive troubleshooting to proactive failure reduction. The type and extent of the risk analysis depends on the stage of the project in the development timeline. It requires the right mindset, using the right tools and performing the appropriate amount of work at the proper time during the development timeline. Applying the appropriate risk assessments tools at the right time could lead to prevention of method failures, better understanding of the critical risks, communication of risks within the analytical team, across teams (process chemistry, formulation, regulatory and quality), and with potential customers of the method (that is, stability groups or technical operations).

Acknowledgements: Rosario LoBrutto would like to thank his Teva colleagues Bianca Avramovitch, Inna Ben-Anat, and Alexey Makarov for their assistance in reviewing this work.

Rosario LoBrutto is the senior director of the Sterile Product Development Group at TEVA Pharmaceuticals, Pomona, New York, USA. He is an active member of the global analytical leadership team and Analytical QbD champion and also serves on the USP expert council for methods validation and verification.

Rosario has over 18 years of experience in the development of active pharmaceutical ingredients (APIs) and drug products including small molecules, proteins and peptides. Prior to joining Teva, Rosario worked with Novartis Pharmaceuticals and Merck Research Laboratories. At Novartis he was project leader for API/drug products, global QbD network leader, and led the global QbD training programme, global specification setting strategy team and global team for streamlining processes for CMC aspects of small molecules/ biologics development projects for IND/IMPD to NDA/MAA/BLA. Rosario received his Ph.D. in analytical chemistry from Seton Hall University (New Jersey, USA).

*Disclaimer: This above views are solely those of Rosario Lobrutto and not representative of Teva.