Different calculation methods for flow cytometric S-phase fraction: prognostic implications in breast cancer? The...

S-phase fraction (SPF), estimated in the flow cytometric DNA histogram, is a prognostic factor in breast cancer. There are, however, some inherent difficulties in the estimation of SPF, such as the influence of debris, aggregates, and normal cells. Most of the available SPF calculation principles try to consider these difficulties, but so far no consensus has been reached with regard to which principle is to be recommended. The aim of the present study was to investigate the prognostic impact of SPF when estimated with different calculation methods in frozen breast cancer samples from 350 patients. Two nonparametric (Rman, Rmin/both rectangle) and three parametric (ACAS/DNA-base, ModFit, and MultiCycle) calculation methods, with and without correction for debris and aggregates, were used. The mean values for SPF varied from 4.3% (ACAS/DNA-base with correction for debris and aggregates) to 9.4% (MultiCycle without any correction for background). The pairwise correlation between methods varied considerably (R = 0.72-0.98). After categorization of SPF values into low SPF (lower two tertiles) and high SPF (upper tertile), all methods yielded statistically significant Pvalues for recurrence-free survival (median follow-up time 67 months), both univariately (0.0004-< 0.0001) and multivariately (0.048-0.0004), after adjusting for nodal status, tumor size, and estrogen receptor status. SPF with background correction did not yield lower P values than SPF without. Regardless of which method was used, SPF showed similar correlations with lymph node involvement, tumor size, and estrogen receptor content. In conclusion, as the mean value of SPF for different calculation methods varies, each laboratory must be restricted to use only one method. Background correction does not seem to improve the prognostic impact of SPF in DNA histograms. Based on the experiences obtained in the present study, S-phase calculation methods without background correction may therefore be the most suitable for routine evaluation of DNA histograms of fresh frozen breast cancer material (ModFit, MultiCycle, and Rman [the latter only for experienced operators]). The nonparametric Rmin, with an automatic setting of the region used for SPF calculation, may be an alternative, but suffers from the disadvantage of not being commercially available yet.