No abstract

No abstract

Protein A chromatography is widely used as a capture step in monoclonal antibody (mAb) purification processes. Antibodies and Fc fusion proteins can be efficiently purified from the majority of other complex components in harvested cell culture fluid (HCCF). Protein A chromatography is also capable of removing modest levels of viruses and is often validated for viral clearance. Historical data mining of Genentech and FDA/CDER databases systematically evaluated the removal of model viruses by Protein A chromatography. First, we found that for each model virus, removal by Protein A chromatography varies significantly across mAbs, while remains consistent within a specific mAb product, even across the acceptable ranges of the process parameters. In addition, our analysis revealed a correlation between retrovirus and parvovirus removal, with retrovirus data generally possessing a greater clearance factor. Finally, we describe a multivariate approach used to evaluate process parameter impacts on viral clearance, based on the levels of retrovirus-like particles (RVLP) present among process characterization study samples. It was shown that RVLP removal by Protein A is robust, that is, parameter effects were not observed across the ranges tested. Robustness of RVLP removal by Protein A also correlates with that for other model viruses such as X-MuLV, MMV, and SV40. The data supports that evaluating RVLP removal using process characterization study samples can establish multivariate acceptable ranges for virus removal by the protein A step for QbD. By measuring RVLP instead of a model retrovirus, it may alleviate some of the technical and economic challenges associated with performing large, design-of-experiment (DoE)—type virus spiking studies. This approach could also serve to provide useful insight when designing strategies to ensure viral safety in the manufacturing of a biopharmaceutical product.

The quality-by-design (QbD) regulatory initiative promotes the development of process design spaces describing the multidimensional effects and interactions of process variables on critical quality attributes of therapeutic products. However, because of the complex nature of production processes, strategies must be devised to provide for design space development with reasonable allocation of resources while maintaining highly dependable results. Here, we discuss strategies for the determination of design spaces for viral clearance by anion exchange chromatography (AEX) during purification of monoclonal antibodies. We developed a risk assessment for AEX using a formalized method and applying previous knowledge of the effects of certain variables and the mechanism of action for virus removal by this process. We then use design-of-experiments (DOE) concepts to perform a highly fractionated factorial experiment and show that varying many process parameters simultaneously over wide ranges does not affect the ability of the AEX process to remove endogenous retrovirus-like particles from CHO-cell derived feedstocks. Finally, we performed a full factorial design and observed that a high degree of viral clearance was obtained for three different model viruses when the most significant process parameters were varied over ranges relevant to typical manufacturing processes. These experiments indicate the robust nature of viral clearance by the AEX process as well as the design space where removal of viral impurities and contaminants can be assured. In addition, the concepts and methodology presented here provides a general approach for the development of design spaces to assure that quality of biotherapeutic products is maintained.

The primary model for cluster analysis is the latent class model. This model yields the mixture likelihood. Due to numerous local maxima, the success of the EM algorithm in maximizing the mixture likelihood depends on the initial starting point of the algorithm. In this article, good starting points for the EM algorithm are obtained by applying classification methods to randomly selected subsamples of the data. The performance of the resulting two-step algorithm, classification followed by EM, is compared to, and found superior to, the baseline algorithm of EM started from a random partition of the data. Though the algorithm is not complicated, comparing it to the baseline algorithm and assessing its performance with several classification methods is nontrivial. The strategy employed for comparing the algorithms is to identify canonical forms for the easiest and most difficult datasets to cluster within a large collection of cluster datasets and then to compare the performance of the two algorithms on these datasets. This has led to the discovery that, in the case of three homogeneous clusters, the most difficult datasets to cluster are those in which the clusters are arranged on a line and the easiest are those in which the clusters are arranged on an equilateral triangle. The performance of the twostep algorithm is assessed using several classification methods and is shown to be able to cluster large, difficult datasets consisting of three highly overlapping clusters arranged on a line with 10,000 observations and 8 variables.