Enumeration of CD4(+) T-cells in the peripheral blood of HIV-infected patients: an interlaboratory study of the FACSCount system.

The aim of the present study was to assess the interlaboratory reproducibility of the FACSCount system for the enumeration of peripheral blood (PB) CD4(+) T-cells. In each of the seven participating centers, both previously stained and unstained PB samples (n = 49) were received and either analyzed or stained and then analyzed. Interlaboratory reproducibility was checked in two different groups of centers (n = 3 and n = 4) where the study was performed in parallel. In addition, both the intralaboratory precision and accuracy of this system were analyzed in comparison with results obtained with conventional flow cytometry. Accordingly, upon comparing both methods, a high degree of correlation was observed in the total number of CD3(+) T-cells (coefficient of correlation of 0.9750 +/- 0.0184, slope of the best linear fit: 0. 9214 +/- 0.0311, y-intercept of 12 +/- 47) as well as in the number of CD3(+)/CD4(+) (coefficient of correlation of 0.9794 +/- 0.1457, slope of the best linear fit: 0.9463 +/- 0.0753, y-intercept of -11 +/- 36) and CD3(+)/CD8(+) (coefficient of correlation of 0.9728 +/- 0.0192, slope of the best linear fit: 0.9682 +/- 0.0735, y-intercept of 7 +/- 95) major subsets. In addition, low coefficients of variation (CV) were obtained for replicates, indicating the method's high degree of accuracy. The present study shows that with respect to the interlaboratory reproducibility reported for most techniques used for the enumeration of PB CD4(+) T-cells, the FACSCount system results in data with much lower coefficients of variance (CVs) (mean CV of less than 10%). Upon measuring the impact on results of different variables associated with either sample preparation or data acquisition and analysis, our study clearly shows that data acquisition and analysis does not influence the results by increasing variability since the coefficients of variation obtained for samples prepared in the same laboratory under the same conditions and read in different laboratories with different instruments were identical to those obtained for the replicates of the same samples read in each individual center. In contrast, interlaboratory variability, although low, significantly increased when sample preparation was carried out in different laboratories, suggesting that pipetting still represents the major source of variability in the FACSCount system. Copyright 1999 Wiley-Liss, Inc.