Evaluation of an alternative to the Kolmogorov–Smirnov test for flow cytometric histogram comparisons

Abstract

Background: The Kolmogorov–Smirnov test is a valid statistical test for comparing distributions that has been recommended for flow cytometric histogram analysis. However, this test is frequently found to be too sensitive for flow cytometric histogram comparisons. Here, a parametric alternative to the Kolmogorov–Smirnov test is proposed that is based on fitting suitable models to flow cytometric data. Methods: Several flow cytometric histograms derived from cell surface immunophenotyping for intercellular adhesion molecule-1 (ICAM-1) on K562 cells were analyzed using numerical modeling. The prediction intervals derived from the modeling were used for decision making. Results: The residuals after peak fitting flow cytometric data are normally distributed and this permits the use of the prediction limit methodology. The usefulness of the approach for analyzing flow cytometry histograms is examined and the method is shown to avoid the “sensitivity” disadvantages associated with the Kolmogorov–Smirnov test. Conclusions: The prediction limit method is a viable alternative to the Kolmogorov–Smirnov method.