(759g) Improving Yield of Drug Product Manufacturing with Hybrid Approach | AIChE

(759g) Improving Yield of Drug Product Manufacturing with Hybrid Approach

Authors 

Eberle, L. - Presenter, Swiss Federal Institute of Technology, Zurich (ETHZ)
Sugiyama, H. - Presenter, The University of Tokyo
Papadokonstantakis, S. - Presenter, Swiss Federal Institute of Technology, Zurich (ETHZ)
Graser, A. - Presenter, F.Hoffmann-La Roche
Schmidt, R. - Presenter, F.Hoffmann-La Roche
Hungerbuehler, K. - Presenter, Swiss Federal Institute of Technology, Zurich (ETHZ)

For remaining competitive in times of a growing public pressure on drug prices and the shrinking research productivity in the industry, drug product manufacturing must be performed increasingly efficient. The presented work is an expansion of Eberle et al. (2014), which comprises of a four-level approach for yield enhancements and is applied in an industrial case study to the manufacturing of injectable drug products, also referred to as Parenterals.

The four levels of the presented hybrid approach combine heuristic and statistical, data-based elements to take advantage of process specific information both from expert knowledge as well as process data. On the first level, the set of loss causing production processes at a unit under review are listed before identifying the processes with most improvement potential. Then, on the second level, expert knowledge is consulted for generating loss scenarios and assigning measurable parameters. Descriptive statistical modelling is performed on the third level before deriving enhancements or escalating process investigations on the fourth level.

As a case study, the presented approach was applied on an annual production output of a manufacturing facility for injectable drug products of Roche in Switzerland. The sampled output comprises of six products from liquid vials and pre-filled batches. Identified costs for lost product units on the first level revealed, that about 20 % of costs for non-commercialized products result from sampling for quality control activities. This loss cause was followed in magnitude by two different glass container defects, i.e. imperfectly formed and scratched vials. Scratched vials and their possible generation during filling processes were modelled with a variety of explanatory variables. However, the statistical models on these variables did not lead to the derivation of implementable information and, as a consequence of that, to an escalation of process investigations. An investigation of scratches with a vial replica that includes a pressure-sensitive outer surface was initiated; enabling the quantification of physical exposures on the glass body. Based on the resulting exposure information, processing steps were prioritised by their impact on glass scratches, hot spots were diagnosed and the statistical model was refined. In summary, the presented data-based approach structured the use of production data and supported efficiency enhancements of production processes.

References: Eberle L., Sugiyama H., Papadokonstantakis S., Graser A., Schmidt R., Hungerbühler K., 2014, Data-Based Tiered Approach for Improving Pharmaceutical Batch Production Processes, Proceedings of the 24th European Symposium on Computer Aided Process Engineering, Elsevier, Amsterdam, the Netherlands.