Developing an expert system for immunophenotypical diagnosis in immunodeficiency. Age-related reference values of...

In the process of developing a decision support system based on flow cytometric data for the diagnosis of immunodeficiency, assessment of lymphocyte subpopulations in human peripheral blood provides the key for further analysis. Samples from 273 'healthy' Hungarian subjects were measured between 1998 and 1999. Immunophenotypic data are compared here (unadjusted for gender) by different age groups: I 0-6 years (n=45); II 7-18 years (n=71); III 19-35 years (n=72); IV 36-55 years (n=48); and V 56-99 years (n=37). Two-color flow cytometric analysis was performed using the Becton Dickinson Simultest IMK Plus kit (CD45/CD14, isotype control, CD3/CD19, CD4/CD8, CD3/HLA-DR, CD3/CD16+56). All lymphocyte subpopulations were measured in all blood samples identically. The quality criteria involved at least 95% of total lymphocytes in the analysis gate, homogenous CD45+ lymphocyte population (minimum of 2000 events in the gate, CD45+ >95%). The frequency of B lymphocytes was the highest, significantly, in the youngest Hungarian subjects, but there were no significant changes with age comparing the data of other II-V age groups. On the other hand, some T subpopulations changed with aging; both CD4 and CD8 subsets varied over time including the elevation of the fraction of activated T cells as well as LGL-NK cells. Some of these changes were significant by statistical tests. Interpretation of flow cytometric data is time-consuming and requires human knowledge of an expert laboratory staff. To facilitate the diagnosis of immunodeficiency, a pilot study aiming at the development of a diagnostic algorithm has been initiated. Algorithm nodes compare the frequency of each lymphocyte subpopulations to the generated reference values. This knowledge-based system describes a short summary report as a result of the comparison, and points to some values requiring further human examination to reach a final conclusion. These reference values and the expert system appear to be a recommended basis for comparing and combining results from different laboratories.