The Path to Digital CPV

In the field of pharma and biopharma, data is currently underutilized. In fact, almost 70% of all manufacturing data collected either is not actionable or the quality of data is insufficient (Manzano & Langer, 2018). To optimize their processes, many manufacturers find advantages of using the available data to gain insights about processes and operations. Collecting and gathering the right process and quality data, correctly storing it, and extracting relevant information in real-time, keep us in the right path to implement a seamless digital Continued Process Verification (CPV).

The process validation lifecycle encompasses three stages: Process Development (Stage 1); Process Qualification (Stage 2) and Continued Process Verification (Stage 3, CPV). The main objective of CPV is to detect variability in the process that may not have been evident when the process was first designed and introduced, particularly in Stage 1. By continuously verifying and adapting the control strategy of the process, previously defined during Stages 1 and 2, the desired product quality is kept throughout its lifecycle, while process knowledge increases. Furthermore, this process can identify opportunities to improve performance or optimize process control (Adhibhatta et al., 2017). When thinking about the digitalization of this whole process, a whole dimension of possibilities is unraveled. The combination of systematic data collection and real-time trend analysis equip the user to proactively identify out-of-trend (OOT) results, which can be used to trigger alarms well before it becomes a major product quality, such as an out-of-specification (OOS). The alarms to the user create a feedback loop to the process making it more dynamic.

All these improvements render the control strategy more robust and reactive, triggering the user to effectively adapt the process to the current conditions, increasing process efficiency and product quality.

Moreover, these new insights can act as a source of evidence to re-evaluate initial risk assumptions made at Stage 1 and Stage 2.

Furthermore, ICH Q8(R2), establishes the grounds for process development under the paradigm of Quality by Design (QbD) and where there should be a strong reliance on PAT for the real-time measurement of CQA. A CPV for a process that is developed under ICH Q8(R2) principles will benefit from a dataset and process knowledge is key in developing a more robust CPV program.

In parallel, ICH Q9(R1) introduces the Quality Risk Management (QRM) framework. By relying on the CPV process to support the criticality assessment, the CPV itself provides the evidence and metrics for risk review, bridging these two concepts (International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use, 2009, 2023). The value of continuously refining risks based on evidence, as part of the on-going process verification, is enormous. The gains are easily picturable: increased process knowledge and increased process control, which consequently reduce OOTs, OOS, financial losses, and compliance risks by approaching the process lifecycle with a preventive mindset anchored in robust digital strategies (Silva, R. C., et al., 2023).

Areas of Revision

Starting with FDA’s guidance “Process Validation: General Principles and Practices” released in 2011 (U.S. FDA, 2011), there are several considerations to be considered when implementing a CPV plan:

• The data collected should capture relevant process trends and contain critical information on the quality of both inputs and the final output.
• Data should be statistically trended and reviewed by trained personnel.
• Quality attributes should be appropriately monitored and controlled throughout the process.

The gains of CPV implementation spread much more that only fulfilling the regulatory agencies requirements. By collecting and analyzing critical data in real-time, it is possible to identify sources of variability and take control over them, which results in a more complete control strategy, more process robustness and, in the end, more process knowledge. In this scenario, risk analysis is streamlined, and mitigation plans are easy to put in place. All these factors contribute to ensuring continued market supply and establishing a solid knowledge base for implementing improvements over the product lifecycle. The changes that come along the way are easier to implement due to the higher level of trust from regulatory authorities. With a fully digital CPV the possibility of achieving all these goals is just around the corner. Leaving all manual processes behind is the only way to extract the full potential of this framework.

Where are the challenges?

With all these gains come a lot of challenges, which depend on the stage of the product within its lifecycle. For instance, if we have a new product, there is a very reduced amount or complete lack of historical data and the level of knowledge about the process and the product is still very limited. The implementation of the Quality by Design (QbD) framework is going to be based on other similar products or prior experience. All process and product knowledge are going to be created during the product development phase. A CPV plan created without a significant amount of process or product knowledge will need periodic reviews and adjustments when more data becomes available.

On another level, for legacy products, a high volume of process and product data is already available. Despite the amount of available data, a significant part of it is not relevant at all. This can create extra noise and/or lead to wrong conclusions. Also, all this data is spread across several formats and platforms. The main challenge is how to automatically assemble and integrate all the data and select the meaningful data subset to include in the CPV plan.

Digital platforms can help with CPV implementation at any stage of the product lifecycle. The reality is that, since most of the data is spread across several different systems, which act as silos, and there is still a lot of data stored in paper format or into spreadsheets, it makes the fully automated integration an unreachable dream. For that reason, a digital solution that provides data intake templates to load historical data into a platform whether it is loaded as a spreadsheet format or inserted as a single data point is mandatory for digital CPV implementation. The first step to digital transition is to leave the paper behind once and for all. By directly feeding the digital solution, data integrity problems due to dependency on manual activities are avoided. For that reason, before stepping into the CPV implementation, all data sources need to be mapped to ensure that the digital solution that is enabling your CPV receives all the necessary process and quality data. It is essential to have all the different systems connected to a single platform to extract, load and transform all the data and be able to produce the correct visualizations for a workflow like CPV.

To ensure compliance with data regulations described in ALCOA+, FDA 21 CFR Part 11, and GAMP 5 (U.S. FDA, 2018), (U.S. FDA, 2003), (ISPE, 2022), the CPV system needs specific requirements. All inputs, as well as data from different sources, need proper validation. Data must not be duplicated and must always be backed up. Furthermore, the access to all data should be controlled and all actions and plans should go through an audit trail. This is the only way to be digitally mature and, therefore, compliant.

The digital CPV workflow

To set up a CPV program that is both reliable, compliant, explainable, and scalable to the rest of your company, a digital workflow approach is the way to go. This workflow is comprised into three steps: critical variable definition, reference batch selection and variable configuration.

Critical variable definition

It is mandatory to include the Critical Quality Attributes (CQA) and Critical Process Parameters (CPP) that result from the risk analysis of your process in the CPV Plan. In terms of digitalization, it is recommended the use of a risk management platform to perform this risk analysis. Ideally, this platform will allow to make the risk assessment based in information but also supported by data.

Reference batch selection

A multivariate approach should be used to select the best and most representative batches. It is important to account for the system variabilities of the CQAs, CPPs and Critical Material Attributes (CMA). By doing this, we are making sure that the starting statistics are solid and representative of the whole process, giving a pre-defined state of control. The obtained quality design space will serve as a guide to evaluate potential process deviations during routine manufacturing.

Variable configuration

The starting point of this step is to create Statistical Control Charts for each CQA, CPP and CMA. There are other aspects to take into consideration at this phase:

• What approaches should be taken to reach a normal distribution.
• What metrics to be included (average, standard deviation, specification limits, etc.)
• What types of Statistical Control Charts (Shewhart, cumulative sum, etc.)

Besides the Statistical Control Charts, it is possible to monitor using other metrics within a CPV plan. For example, metrics related to process capability, such as the Process Capability Index (CpK) or the Process Performance Index (PpK).

Finally, when using an online platform for the CPV, alarms based on rules to help speed up decision-making can be defined. By building a decision tree that triggers respective alarms, the whole process can be monitored and the responsible can be notified in real-time whenever there is a statistical deviation.

What are some of the benefits of a complete CPV solution?

First, it is important to keep in mind that CPV is all about delivering a much higher product quality assurance and bringing excellence to your operations.

An Online CPV tool has many benefits when compared to an offline counterpart. Some of the benefits of a complete online CPV tool are:

1. Data Integration - Connection between different systems to integrate both process and quality data, according to the needs of your CPV Plan.

2. Data Acquisition - Automatic data collection is the only way to have data integrity across the full process.

3. Automatic metric updates - The system will automatically update the process capability metrics as well as the control charts whenever there is new data.

4. Alarm triggers - The system will react and alert if the process performance differs from expectation.

5. Reporting - Automatic report creation that meets the pre-defined requirements and that can be used for regulatory QA usage, fully aligned with your PQS systems.

In addition to providing a continuous collection and analysis of critical data, a CPV program allows for a more efficient and complete control strategy. It will help identifying sources of variability and be more effective at their monitoring. It will result in a more robust and predictable process that creates a higher level of trust near the regulatory authorities during inspections and facilitates building a compelling argument for the introduction of post-approval changes.

Take-away messages

As stated, a manual CPV plan is a time-consuming process. And one of the main reasons is because it is a reactive process. With a manual CPV, most likely you only know that there was a problem in your process after it has occurred. That’s the main reason to level up for a digital CPV approach. It is automatic, predictive and responsive. Therefore, you’ll have the time to act on process deviations and improvements before they become serious product quality problems.

Companies will largely benefit from the value of continuously refining risks based on evidence, as part of the on-going process verification. Combining process and risk data in a retrofitting loop is the key to make informed decisions on your process, reducing the subjectivity of quality decisions and resulting in a solid knowledge base for the implementation of improvements throughout the product’s lifecycle.

References

1. Manzano, T., & Langer, G. (2018). Getting ready for pharma 4.0. Data integrity in cloud and big data applications. Pharmaceutical Engineering, 72–70. https://www.ispe.gr.jp/ISPE/02_katsudou/pdf/201812_en.pdf

2. Adhibhatta, S., DiMartino, M., Falcon, R., Haman, E., Legg, K., Payne, R., Pipkins, K., & Zamamiri, A. (2017). Continued Process Verification (CPV) Signal Responses in Biopharma. Pharmaceutical Engineering. https://ispe.org/pharmaceutical-engineering/january-february-2017/continued-process-verification-cpv-signal

3. U.S. FDA. (2011). Process Validation: General Principles and Practices. https://www.fda.gov/files/drugs/published/Process-Validation--General-Principles-and-Practices.pdf

4. International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use. (2009). ICH guideline Q8 (R2) on pharmaceutical development. https://www.ema.europa.eu/en/documents/scientific-guideline/international-conference-harmonisation-technical-requirements-registration-pharmaceuticals-human-use_en-11.pdf

5. International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use. (2023). Quality Risk Management Q9(R1). https://database.ich.org/sites/default/files/ICH_Q9%28R1%29_Guideline_Step4_2023_0126_0.pdf

6. U.S. FDA. (2003). Part 11, Electronic Records; Electronic Signatures – Scope and Application. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/part-11-electronic-records-electronic-signatures-scope-and-application

7. U.S. FDA. (2018). Data Integrity and Compliance With Drug CGMP: Questions and Answers. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/data-integrity-and-compliance-drug-cgmp-questions-and-answers

8. ISPE. (2022). GAMP 5 Guide (2nd ed.).

9. Silva, R. C., Almeida, R., Ferreira, P., Menezes, J., & Martinho, Â. (2023, July). Agile, Data-Driven Life Cycle Management for continuous manufacturing. Pharmaceutical Engineering. https://ispe.org/pharmaceutical-engineering/july-august-2023/agile-data-driven-life-cycle-management-continuous