Comparison of five cell cycle analysis models applied to 1414 flow cytometric DNA histograms of fresh frozen breast cancer

Conflicting prognostic results with regard to DNA flow cytometric variables have been reported for breast cancer patients. Reasons for this can be found mainly on the different levels of methodology, including the interpretation of the DNA-histograms. Several computer programs based on different fitting models are available for cell cycle analyses which result in different %S-phase calculations. The present study evaluated the influence of 5 different cell cycle analysis models on several cell cycle variables (%S-phase, %G2M-phase, %diploid cells, DNA-index, %debris) derived from flow cytometric DNA-histograms obtained from breast cancers. DNA-histograms obtained from 1414 fresh frozen breast cancers were interpreted using 5 different cell cycle analysis models using the computer program MultiCycle AV. Model 1 used the zero order S-phase calculation and 'sliced nuclei' debris correction, model 2 added fixed G0/G1 and G2/M-phase ratio, and model 3 added correction for aggregates. Model 4 applied the first order S-phase calculation and sliced nuclei debris correction. Model 5 fixed the CVs of the G0/G1 and G2/M-phase in addition to applying the sliced nuclei debris correction and zero-order S-phase calculation. Using all cases, it was shown that when the aggregates correction was included (model 3) in the analysis, on average, significantly lower mean values were obtained for %S-phase cells, and %debris, and %G2M-phase cells of the first cell cycle. No significant differences were observed for the other variables. Analyzing the DNA-diploid, tetraploid, and aneuploid cases separately, similar results were obtained. Linear regression analysis showed only moderately strong correlations for the %S-phase and %G2M-phase variables between the different models, indicating that for individual DNA-histograms the cell cycle analysis results may vary. In conclusion, quite different values can be obtained for especially the %S-phase cells using different cell cycle analysis models in individual cases. Correction for aggregates results on average in significantly lower %S-phase values. This clearly has implications for comparing %S-phase results from studies using aggregate correction or not, especially with regard to prognostic thresholds. Large follow-up studies are necessary to derive at the prognostically best model.