Using Data Analytics to Fast-track Chromatography Development

Data analytics using MODDE^® Design of Experiments (DOE) helps you plan, conduct and analyze chromatography formats, including Ion Exchange (IEX), Protein A, and membrane, to optimize downstream process development. DOE helps uncover multiparameter relationships, minimize impurities, maximize yield, and improve downstream economy.

^{This article is posted on our Science Snippets Blog}

For biopharmaceutical companies that rely on chromatography for downstream process development, Design of Experiments (DOE) is a useful tool because it helps you reduce the number of trial-and-error experiments needed to establish a robust process for your product, and at the same time, develop a process that provides the best yield, in the shortest time, with the least amount of waste.

According to Sartorius Data Scientist, Gregory Case, “taking a data analytics approach to chromatography development is especially important, because if you ´just wing it`, you’ll miss the interactions that multiple variables can have on each other”. For example, conditions that maximize yield could have worse clearance for purification.

On the other hand, a DOE approach helps you identify the optimal conditions that meet all your objectives with fewer experiments, and without missing something you might not have considered otherwise.

DOE is used to gain knowledge, increase process understanding, and establish optimal process conditions (“design space estimation”).

Advantages of Using Data Analytics for Bioprocess Development

What are the advantages of using Design of Experiments (DOE) for bioprocess development?

• You can review multiple factors at once
  DOE helps you create a more robust process that takes into account multiparameter relationships. A typical approach to process development, including using chromatography, involves changing variables one at a time to observe the effect they have on yield or quality. For example, you might adjust the load, pH, or conductivity in a column.
  However, DOE is not limited to looking at just two factors. It can be applied to three, four or many more factors at time, and take into account the interactions that happen between them. It helps you uncover things you might not notice if you conduct experiments one at a time. For example, if you adjusted pH or load individually, you might think one improves yield and another quality. But at one point do the two interact, or have an optimal level? DOE gives you information you would miss otherwise.
• You can apply DOE early in process development
  Sometimes scientists think that DOE doesn’t apply to their process, or that they are too early in the process development for DOE to be effective. But the opposite is actually true. Using data analytics from the start helps create more robust processes that can be readily quantified and scaled up. DOE helps you validate and characterize process stability and robustness and identify critical process parameters (CPPs) that may be sensitive to small changes.
• DOE supports a Quality by Design (QbD) approach
  Data Analytics supports a Quality by Design (QbD) approach to process development that is favored by regulatory agencies. In addition to creating more robust formulas and well-documented processes, following a QbD approach can help ease regulatory compliance reviews and ensure a stable process for long-term production optimization. Using QbD approach early in your process development helps set you up for regulatory approvals and more efficient process scale-up later.
• Helps accelerate process development and scale-up
  Data analytics helps accelerate downstream process development and scale-up. Companies have decreased their experimental time or number of iterations needed by 50% using DOE.
• Sets the stage for real-time analytics
  Using DOE helps you establish a culture based on data analytics in which real-time monitoring and real-time analysis can be applied.  This can be essential for digital transformation and supports a regulatory-favored approach based on QbD.
• Use wizards to support design space selection
  If you use MODDE^® Design of Experiments software, you have access to wizards that makes it easy to set up DOE in an efficient way that translates into a QbD approach.

Three Examples: How DOE Supports Chromatography Studies

Consider these three examples that showcase how data analytics be used to more effectively plan, conduct, and analyze chromatography  development studies (including Ion Exchange (IEX), Protein A, membrane chromatography).
 

1. Develop Membrane Chromatography Schemes for antibody-drug conjugate (ADC) Clearance
During antibody downstream processing, membrane chromatography is routinely used to remove host cell proteins, viral particles, and aggregates.  The application of membrane chromatography for ADCs manufacturing has been applied in a limited capacity and in only specialized scenarios. ^[1]

Abzena used MODDE^® | DOE to develop and optimize binding, washing, and elution conditions for single-step membrane chromatography. The benefits realized were:

• Systematically identify multiparameter relationships
• Improve chromatography performance
• Decrease aggregates and increase drug antibody ratio (DAR) homogeneity

^{Two factors were defined (buffer A concentration, buffer B concentration) and two responses (DAR, yield). The Design Space plot shows the probability of failure (%) for factor combinations.}

Download the Poster
 

2. Optimize Ion Exchange (IEX) Chromatography Loading Conditions

More than 60% of scientists^[2] say that high resolution or purity is their top priority or challenge in IEX. Balancing product yield against product purity is major challenge in chromatography loading optimization studies and chromatography performance can depend on many interdependent process factors.

Cytivia used MODDE^® | DOE to plan, conduct, and analyze chromatography optimization studies. The benefit realized were:

• Systematically identify multiparameter relationships
• Minimize impurities while maximizing yield
• Simplify IEX chromatography development

The experimental setup defined 3 factors (load, pH, and conductivity) and 4 responses (yield, host cell protein (HCP) clearance, dimers and aggregates (D/A), and protein A clearance.
The results?

• Yield: Load and pH have a strong impact
• HCP Clearance: Load, pH and conductivity have a strong impact
• D/A Clearance: pH has a strong impact
• Protein A Clearance: pH has a strong impact

How to find the optimal settings for each of these parameters?

The response contour plots show the influence of sample load, pH, and conductivity on the four responses, and how to reach desired values for each of them. The interactions between each of the factors were accounted for using data analytics.

• Yield: Optimized at higher sample loads and low pH
• HCP Clearance: Optimized at high conductivity and low pH
• D/A Clearance: Optimized at high load and low pH
• Protein A Clearance: Optimized at high conductivity and low pH

^{Suggested loading conditions: sample load of 200mg/mL, pH 7, conductivity 8.5mS/cm; Expected outcome:  yield > 90%, leached Protein A below detection limit, D/A < 0.5%, and HCP concentration of < 15ppm}

3.  Optimize Protein A Chromatography Loading Time

Balancing product yield against product purity is a major challenge in chromatography loading optimization studies. Optimizing residence time in chromatography systems has the potential to reduce dynamic capacity and increase overall productivity, however, this can depend on many interdependent process factors.

Cytiva and Takeda used MODDE^® | DOE to plan, conduct, and analyze chromatography optimization studies. The benefit realized were:

• Systematically identify multiparameter relationships
• Minimize impurities while maximizing yield
• Optimize chromatography economy

How to design an innovative loading strategy to improve productivity and economy?

The goal of this experiment was to investigate the possibility to optimize residence time (RT) during loading to improve productivity without significantly affecting capacity. RT is defined as the time taken for a solute to pass through a chromatography column.

^{One way to address this is to use variable RTs.}

How to find the best combination of RTs to maximize productivity and minimize load time?

The experimental set-up included two factors (RT1 and RT2) and two responses (productivity increase and load time saved). Ten experiments were performed using variable loading times RT1 (1.2 to 2min) and RT2 (2.5 to 4.5min). Reference RT3 was kept constant at 6 minutes. The results? An approximately 40% productivity increase and approximately 55% load time saved when using variable loads of 1.6, 3.5, and 6 minutes.  Interacting factors that would not have been seen without DOE.

Conclusion

Data analytics is a useful tool in downstream process development because it helps discover multiparameter relationships, minimize impurities, maximize yield, and improve downstream economy.  Additionally, building a data analytics culture and way of working from an early stage will enable conditions for rapid scale-up from clinical to manufacturing – where real-time analytics can be applied.

   + Read more          Why DOE Is Essential in the (Bio)Pharma Industry

   + Watch webinar        Watch Recorded Webinar: Robust Optimization Made Easy

   + Learn more        Learn when and how to use DOE

^Reference

^{1.    Cordova JC, Sun S, Bos J, Thirumalairajan S, Ghone S, Hirai M, Busse RA, Hardt JSvd, Schwartz I, Zhou J. Development of a Single-Step Antibody–Drug Conjugate Purification Process with Membrane Chromatography. Journal of Clinical Medicine. 2021, 10(3):552.
2.    Cytiva. Ion Exhange Chromatography, Top 4 Tips to Achieve High Protein Purity.2020
3.    Ericson A, Antti M, Johansson H. Cytiva Lifesciences Application Note 28-9078-89 AA. Optimization of Loading Conditions on Capto adhere Design of Experiments., 2020.
4.    Björkman T. Cytiva Lifesciences Application Note, 29190587 AA. Optimize column loading strategy to gain productivity in protein purification. 2016}