Application of Design of Experiment (DOE) Principles to the Development of Biologic Control Materials in Immunohistochemistry
Julie Erickson, Dennis Huang, Marsha Hudson, Scott Webster. Dako North America, Inc., Carpinteria, CA
Background: Cell lines with specific antigen expression levels are useful controls for qualifying immunohistochemical (IHC) staining runs. To successfully fulfill this function embedded cell pellets must exhibit the desired level of antigen expression and do so in a robust and reproducible manner. The use of DOE methods and 2k factorial experiments supported by image analysis is an efficient and objective way to optimize the manufacture of these control materials.
Design: To illustrate the DOE approach, key process variables were examined for their effects on the IHC staining intensity of paraffin-embedded HT-29 (colorectal adenocarcinoma) cultured cells. Section thickness and humidity in the drying oven were analyzed in a 22 factorial design, and as a more complex example fixation and dehydration times were examined with embedding matrix type in a 23 factorial experiment. Relative antigen expression levels were estimated using the ACIS® III slide scanner (Dako) and an image analysis algorithm based on staining intensity. Data were evaluated using Minitab® statistical software.
Results: The 22 experiment demonstrated decreased staining intensity with higher humidity (p=0.001) and a nonsignificant trend with thinner section thickness (p=0.152). Higher humidity was also associated with increased variability in intensity and compromised cell-to-slide adherence. The 23 experiment produced significant increases in staining intensity, in this case with increased fixation time (p=0.039) and the use of Type 2 embedding matrix (p=0.000) but not with change in dehydration time. Interestingly, there were two significant interactions between variables. Decreased fixation time lowered staining intensity with the use of Type 1 but not Type 2 embedding matrix (p=0.032) and increased fixation increased intensity with shorter dehydration periods but not with longer periods. Thus, DOE methods effectively reveal the impact of process variables (alone and in combination) on both the magnitude and variation of the response variable.
Conclusions: DOE and factorial experiments provide a systematic way of analyzing processes and optimizing key parameters to achieve a specified result, i.e. a robust and reproducible manufacturing protocol. The use of image analysis complements the DOE approach as it provides an objective scoring method and produces parametric data that are amenable to statistical analysis. It is important to note that each cell line control should be investigated individually as the effects of the manufacturing processes may be assay-specific.
Monday, March 19, 2012 1:00 PM
Poster Session II # 307, Monday Afternoon