Statistical considerations in DNA flow cytometry

Abstract

Flow cytometry is frequently used for assessing individual cell characteristic(s) for a large number of cells. It has a variety of medical applications including assessing the quantity of intracellular DNA and detecting the presence of antigens such as CD4. Flow cytometric variables are evaluated for their clinical prognostic value, particularly in cancer, and are often used for clinical screening of diseased patients. The prognostic worth of these variables is questionable and is controversial in the medical community. This controversy is caused in part by the multiple methods of analysis and the lack of adherence to quality control standards. The analysis of flow cytometric data presents a number of interesting statistical problems, particularly in deconvolution of overlapping distributions and detection of abnormal subpopulation(s) of cells. The current methods incorporate subjective procedures, may use ill-founded assumptions and yield differing results. This article summarizes the flow cytometry process of measurement and reviews unsolved statistical and quality control issues pertaining to the analysis of flow cytometric data. DNA histogram analysis is used to exemplify these issues.