화학공학소재연구정보센터
Biotechnology and Bioengineering, Vol.110, No.10, 2633-2642, 2013
Evaluating Differences of Metabolic Performances: Statistical Methods and Their Application to Animal Cell Cultivations
In cell culture process development, monitoring and analyzing metabolic key parameters is routinely applied to demonstrate specific advantages of one experimental setup over another. It is of great importance that the observed differences and expected improvements are practically relevant and statistically significant. However, a systematic assessment whether observed differences in metabolic rates are statistically significant or not is often missing. This can lead to time-consuming and costly changes of an established biotechnological process due to false positive results. In the present work we demonstrate how well-established statistical tools can be employed to analyze systematically different sources of variations in metabolic rate determinations and to assess, in an unbiased way, their implications on the significance of the observed differences. As a case study, we evaluate differing growth characteristics and metabolic rates of the avian designer cell line AGE1.CR.pIX cultivated in a stirred tank reactor and in a wave bioreactor. Although large differences in metabolic rates and cell growth were expected (due to different aeration, agitation, pH-control, etc.) and partially observed (up to 79%), our results show that the inter-experimental variance between experiments performed under identical conditions but with different pre-cultures is a major contributor to the overall variance of metabolic rates. The lower bounds of the overall relative standard deviations for specific metabolic rates were between 4% and 73%. The application of available statistical methods revealed that the observed differences were statistically not significant and consequently insufficient to confirm relevant differences between both cultivation systems. Our study provides a general guideline for statistical analyses in comparative cultivation studies and emphasizes the necessity to account for the inter-experimental variance (mainly caused by biological variation) to avoid false-positive results. Biotechnol. Bioeng. 2013;110: 2633-2642. (c) 2013 Wiley Periodicals, Inc.