When DoE fails: Mechanistic modeling for chromatographic manufacturability

Antibody purification processes are typically developed on the basis of platform approaches. Optimal operating conditions for single units are found by altering just a few process parameters with Design-of-Experiment (DoE) procedures or high-throughput screenings.

DESIGN-OF-EXPERIMENT FAILED TO IDENTIFY ROBUST PROCESS CONDITIONS

An industrial separation problem of a polishing step using two-step elution was considered. The challenge was to separate an antibody from a mixture containing critical aggregates and fragments, while keeping the adsorber material as well as maintaining high yield and purity. A process development approach based on a platform process in combination with DoE failed to find any suitable process with a yield greater than 65%.

Protein in blue setting

CHOOSING OPTIMIZATION PARAMETERS USING MECHANISTIC KNOWLEDGE

Afterwards, the process was optimized using a model-based approach. For the initiation of process optimization, parameters and a respective objective function had to be chosen. In order to achieve this, an understanding of mechanistic processes was utilized: In the present case, an antibody and its fragments co-eluted despite a difference in surface charge. Because of thermodynamic effects, the more highly-concentrated antibody peak overtook the fragment peak while moving through the column. Given this background, the volume of protein load as well as the column length were considered as additional optimization parameters.

SUCCESSFUL SEPARATION OF TARGET PROTEIN

With the optimized protein load and column length, the mixture could be separated successfully in a two-step elution process. The achieved purity was now above 98% and the yield was 95%. Apart from the given experimental results of the previous DoE, no additional calibration experiments were needed to achieve these results. ChromX enables the variation process parameters and optimization objectives while saving time and costs. As a side effect, mechanistic process understanding was generated and successfully employed for process development.

This publications was presented at the following conferences:

  • 30th International Symposium on Preparative and Process Chromatography (PREP); 07/2017; Philadelphia
  • 253rd ACS National Meeting; 04/2017; San Francsisco

See full poster “When DoE fails: Mechanistic modeling for chromatographic manufacturability

Do NOT follow this link or you will be banned from the site!