Cell culture monitoring: is Raman monitoring for spectroscopy experts only?

Raman spectroscopy has been demonstrated as a powerful technique for in-line cell culture monitoring. However, can we expect an efficient PAT tool leveraging this technique that would be easy-to-use for a non-expert?

Over the past recent years, about ten papers mostly from PAT or process development teams from the biopharmaceutical industry have presented accomplished results. Nevertheless it is clear that the implementation of such pilot tests have been carried out by chemometrics and Raman spectroscopy experts in the framework of big and expensive projects. This work has been very useful in terms of technology assessment. Enhanced cell culture yields by applying a fed-batch monitored by Raman spectroscopy have been obtained. They have pointed out the need for adjustment of data processing for training a chemometrics model. They have described complex interfaces and data management for such an implementation in bioprocess monitoring.

Now the question is: can we dramatically limit the recurrent cost of implementing such a solution for each bioprocess?

Such a complexity is mainly due to the current approach consisting in using “generic” tools (instruments and software suites) that have been designed for Raman spectroscopy or spectroscopy in general, for applications as large as industrial gas manufacturing or the agro-food industry. The effort to customize an analyzer and a data processing flow for monitoring cell cultures is huge, whereas in fact the variety of cases to cover is quite limited (type of Raman spectra, sequencing, data to reject, content of the cell culture medium to be analyzed, compounds to monitor etc.).image1

At RESOLUTION Spectra Systems, we believe and prove it with ProCellics™, that a Raman analyzer, including its software suite, specifically designed for bioprocesses monitoring limits the complexity of the implementation in such a way that it can be used without specific skills in Raman spectroscopy and with limited notions of chemometrics. Building a chemometrics model and testing its robustness can be similar to calibrating an industrial sensor provided that:

  • the user is guided step by step in the process of building the model, consistently with Quality-by-Design
  • the selection of data to be used for building the model is mainly automatic or based on very simple criteria
  • the parameters for building the models are pre-defined and the robustness of the resulting models are tested by the PAT software suite
  • the data are managed according to CFR 21 part 11 which is also the guarantee of an efficient pipeline of data.

Christophe BONNEVILLE

President of RESOLUTION Spectra Systems