Does agile Quality by Design save time or cost?

For an early-stage biotech team, deciding what to investigate next can be difficult. Several ingredients or manufacturing conditions may affect a product, while the available evidence gives only part of the picture.

Quality by Design (QbD) provides a systematic way to investigate those questions. It starts with the quality a product needs to achieve, then builds an understanding of the factors that affect it. An agile approach organizes this work into focused cycles, with each set of findings informing the next decision.

Nicole Cohen, PhD, Senior Director of Chemistry, Manufacturing and Controls at VCLS (Voisin Consulting Life Sciences), wrote the regulatory module for an EIT Health learning project developed with Cybernano. She also reviewed its monoclonal-antibody manufacturing case studies.

“The principle is to ask a question”

A related VCLS–Cybernano webinar shows how that principle works through a teaching example about antibody production.

What does “agile” mean in Quality by Design?

In the approach presented by Cybernano, a team breaks a development problem into questions it can investigate through focused work cycles, called sprints.¹

Each sprint starts by reviewing the product, the manufacturing process and the knowledge already available. The team assesses which uncertainties matter to the question. Where further data are needed, it designs and carries out experiments, then analyzes the results.

The findings guide a decision. There may be enough evidence to proceed, a need for further work, or grounds to stop pursuing a particular route.

Figure 1

One cycle: from a question to a decision

1. Ask the question

   One development question the team needs to resolve before it can move on.

2. Review what is known

   The product, the manufacturing process and the knowledge already available.

3. Assess the uncertainties

   Which unknowns matter to this question, and how confident the team can be in what it has.

4. Design and run experiments

   Planned combinations of conditions, designed to answer the question with the fewest runs.

5. Analyze and decide

   Enough evidence to proceed, a need for further work, or grounds to stop pursuing that route.

The decision sets the next question. Each cycle’s findings narrow what the following cycle needs to investigate.

Throughout: quality objectives, risk assessment and suitable controls, with regulatory input on which questions matter at this stage of development.

Quality by Design also includes setting quality objectives, assessing risks and developing suitable controls. Experimental design and statistical analysis help the team build the understanding needed to make those decisions.

A teaching example: improving antibody concentration

In the webinar presented by Nicole Cohen and Thierry Bastogne, Co-founder and Chief Scientific Officer of Cybernano, a manufacturing process for a monoclonal antibody is failing to meet a target concentration above 5 grams per liter.

The concentration is called the antibody titer: how much antibody is present in a given volume of the cell culture. The training case uses simulated data to illustrate the method.

The presenters divide the investigation into four questions.

Figure 2

Four questions, one teaching case: antibody concentration above 5 g/L

1. Which inputs might matter?

   Existing knowledge ranks feed, temperature, dissolved oxygen, agitation and other inputs by likely impact and by how confident that assessment is.

2. Which do experiments confirm?

   A screening experiment tests planned combinations. In the simulated case, dissolved oxygen and agitation speed emerge as important.

3. Which combinations meet the target?

   A model built from the next experiments predicts where the concentration target could be met: a region of interest.

4. Does the evidence hold up?

   Further experimental data test the proposed operating region to validate a design space, because a prediction needs experimental support.

Each answer narrows the next question: from several suspected inputs to a supported operating region.

Throughout: the quality the product must achieve stays the reference point. The example covers one aspect of process performance; a product’s wider quality requirements need their own consideration.

1. Which inputs might affect the result?

The process has several possible influences, including the feed supplied to the cells, temperature, dissolved oxygen and agitation speed.

The first step uses existing knowledge to assess which inputs may matter and how confident the team can be in that assessment. This helps identify where experiments could resolve important uncertainties.

2. Which inputs do the experiments identify as important?

The teaching case then uses a screening experiment to investigate potentially important inputs. It tests planned combinations of conditions and analyzes their effects on antibody concentration.

In the simulated example, dissolved oxygen and agitation speed emerge as important factors. The results narrow the questions for the next cycle of work.

3. Which combinations of conditions could meet the target?

Knowing that an input matters does not establish which value to use. The next investigation explores combinations of operating conditions and builds a model to predict where the concentration target could be met.

This produces a region of interest for further investigation. The model gives the team a basis for choosing what to test next.

4. Do further experiments support the prediction?

The next cycle uses additional experimental data to assess a proposed operating region, called a design space in QbD.

This final question is important because a model’s prediction needs experimental support. The example shows how successive questions can build understanding of a process and the conditions under which it performs acceptably.

Where regulatory advice fits

Development experiments should help a company understand its product and explain the choices it makes. Regulatory input can help the team identify which questions matter at its stage of development and how the resulting evidence will support future submissions.

Nicole’s regulatory teaching brings together relevant European and US guidance on Quality by Design. Cybernano contributes the agile method and tools for structuring investigations.

For a company considering the approach, an initial discussion can examine what is already known about the product, which development decision comes next and what expertise is needed to investigate it.

The antibody example illustrates the reasoning involved. Its focus on concentration addresses one aspect of process performance. A product’s wider quality requirements also need appropriate consideration.

Common questions about early Quality by Design

Can Quality by Design start before clinical trials?

Yes. QbD principles can inform early development, including preclinical work. The scope should reflect the knowledge available and the decisions the team needs to make. ICH Q8(R2), the pharmaceutical development guideline, also recognizes the relevance of its principles during clinical development.²

Does agile Quality by Design guarantee time or cost savings?

Not always, but it often helps reduce delays caused by poor decisions made under highly uncertain conditions. Several studies have shown that QbD can reduce development time by up to 40% and consumables usage by up to 50%.^{3, 4} The approach aims to make development work more focused, but its benefits depend on the project and how it is implemented. The webinar’s simulated example demonstrates a method of investigation.

What is a design space?

A design space describes combinations of material attributes and process parameters demonstrated to provide assurance of quality. Establishing it requires supporting evidence. A pharmaceutical applicant proposing a design space submits it for regulatory assessment and approval.^{2, 5}

Explore the webinar and training

The VCLS–Cybernano webinar, Agile Quality by Design: A Practical Guide for Modern Pharmaceutical Development, explains the approach and walks through the antibody teaching case.

EIT Health also offers two online courses:

• Quality by Design: Building Practical Skills for Biomanufacturing and Biotech.
• Quality by Design: Expanding Expertise in AI, Regulation and Continuous Improvement.

Both course pages list VCLS as a partner alongside Cybernano, Immerscio.bio and University Medical Center Utrecht.

BLOG POST BY

Nicole Cohen, PhD

Senior Director, CMC