Cell culture bioassays are perceived as laborious and imprecise even after substantial development efforts. These bioassays are used broadly in the biotechnology industry to measure protein or vaccine products in development and for lot release. Imprecise assays and slow assay development contribute to slow product development. Common practice in cell culture bioassay ignores statistical and regulatory guidance by failing to utilize proper randomization and not accounting for group effects (caused by multi-channel pipettes). In addition, the lack of attention to location effects within the assays combined with simplistic analyses of assay data compromise the precision and efficiency of cell culture bioassays. We propose a modern standardized approach to cell culture bioassay using statistical design and analysis linked to laboratory robots, which will address randomization, location effects, serial dilution, and grouped dilution. The larger program goal is to build a turnkey cell culture bioassay system using statistical software and a laboratory robot. The work described in this proposal will use 1) standardized designs and analysis, 2) linear or nonlinear mixed split- or strip-plot models for bioassay analysis, 3) recent developments in laboratory robotics that make randomization practical (both hardware and software), and 4) new code for modeling serial dilution error (developed under EPSCoR funding, Lansky 2004). The combination of methods, once developed, should reduce the coefficient of variation from high single digits or substantially higher to low single digits, reducing the need for replication by a factor of four to ten or more. Most importantly, by building a turnkey robotic and analysis bioassay system, this level of assay performance will be made broadly available to the biotechnology industry as well as to non-profit researchers.
| Buzas, Jeffrey S; Wager, Carrie G; Lansky, David M (2011) Split-plot designs for robotic serial dilution assays. Biometrics 67:1189-96 |
