6600 Background: Decitabine is a therapy for myelodysplastic syndromes (MDS). Myelosuppression is typical of MDS and a common toxicity of decitabine, and may result in dose modification (reduction and/or delay) but may not indicate progression of MDS. METHODS We conducted a retrospective analysis of a pooled subset of patients (pts) with MDS from 2 decitabine dosing trials: 5-day (Phase II) and 3-day (Phase III). Pts with ³1 dose of decitabine were included. Cox regression analyses were conducted of on-study red blood cell (RBC) or platelet (PLT) dependence, occurrence of myelosuppression, dose modification (reduction or delay), and survival on baseline variables and some time-dependent covariates. RESULTS 182 pts were included. RBC or PLT dependence at baseline was a predictor for dose modification on study (RBC: P=.008, hazard ratio [HR]=2.11; PLT: P=.011, HR=1.98) as was myelosuppression on study (P<.0001, HR=8.08). Pts who had dose modification had a better response (P=.0001, HR=2.27) regardless of RBC baseline dependence status, and pts with dose modifications had better overall survival vs those without dose modifications (median [95% CI] 19.2 [15.2, 21.3] vs 10.3 [7.6, 15.7] mo; P=.0039). Time to dose modification also correlated with response; pts with a shorter time to dose modification had a quicker response (P=.0001, HR=2.27). Cycle number may have affected results; pts with dose reduction/delay or modification received a median of 6 cycles vs 2 for other pts. CONCLUSIONS Pts who develop myelosuppression on treatment will likely require dose modification. However, these pts appear to have a significant clinical response. One hypothesis is that decitabine may remove both abnormal and normal rapidly dividing marrow cells during a susceptible phase of the cell cycle, so myelosuppression may indicate decitabine activity. That pts with a shorter time to dose reduction or delay had a better response supports this hypothesis. Based on this retrospective analysis, myelosuppression and dose reductions or delays on treatment may not necessarily indicate lack of efficacy; conversely, they may indicate a response to treatment.