The present study was undertaken to assess the predictive value of pretherapeutic determinants of ara-C metabolism and proliferative activity of leukemic blasts for early response to antileukemic therapy in the setting of granulocyte-macrophage colony-stimulating factor (GM-CSF)-based priming before and during TAD-9 induction in 36 consecutive patients with de novo acute myeloid leukemia (AML). Ara-C metabolism was assessed by the activities of deoxycytidine kinase (DCK), deoxycytidine deaminase (DCD), DNA polymerase alpha (Poly alpha), and overall polymerase (overall Poly). The fraction of cells in S phase (%S phase) and thymidine kinase (TK) activity were determined as a measure of proliferative activity. Early response to therapy was defined by the percentage of leukemic blasts in the bone marrow 5 to 7 days after completion of TAD-9 with less than 5% signaling an adequate response and greater than 5% indicating an inadequate early reduction, respectively. While neither %S phase, DCK, nor overall Poly activity were predictive for early response, TK and Poly alpha activities were significantly higher for cases with adequate blast cell clearance. The respective median values were for TK 3.8 versus 1.85 pmol/min/mg protein (P = .012), and for Poly alpha 1.9 versus 0.69 pmol/min/mg protein (P = .014). An inverse relation was detected for DCD activity which was significantly lower in responding patients with a median of 0.33 nmol/min/mg protein (range, 0.0 to 29.5) as compared to a median of 5.1 nmol/min/mg protein (range, 0.11 to 8.45) in early nonresponders, (P = .009). Taking the respective median values as arbitrary cut-points for high or low enzyme activities, responders and nonresponders could be discriminated prospectively. Hence, 14 of 16 cases (88%) with DCD activities below the median of 1.56 nmol/min/mg protein responded as compared to only 3 of 14 (22%) patients with higher DCD activities (P = .0004). From the 15 patients with TK activity above the overall median of 3.2 pmol/min/mg protein, 11 cases (73%) achieved an adequate blast cell clearance while only 6 of 17 cases (35%) with lower values responded (P = .035). Similarly, 12 of 15 patients (80%) with high Poly alpha levels (>1.22 pmol/min/mg protein) responded to induction therapy as compared to only 5 of 14 patients (36%) with lower enzyme activities (P = .02). By logistic regression analysis of enzyme activities, DCD activity was found to be the most sensitive parameter to predict an adequate blast cell clearance (P = .032). Activities of DCD and TK were not only associated with initial response but were also found predictive for remission duration. Hence, from 11 patients with low TK levels 8 (73%) relapsed within 1 year, whereas only 2 of 11 (18%) patients with high TK activity experienced a recurrence of their disease (P = .015). Six of 9 (66%) patients with higher than median DCD levels relapsed within 1 year, whereas 10 of 14 patients (71%) with lower DCD levels had a longer remission duration (P = .085). Analysis of DCD gene expression at the mRNA level by a semi-quantitative reverse transcriptase-polymerase chain reaction method showed that a high transcription rate of the DCD gene was associated with high enzyme activities and vice versa. Hence, the observed intraindividual differences in DCD activity are a reflection of differences in gene activity and transcription rate rather than of variants in translation. Although further analyses are needed to elucidate the molecular mechanisms that determine the variation of enzyme activities in individual patients, the present study strongly suggests that pretherapeutic determination of TK and Poly alpha as well as of DCD allows to predict response to TAD-9 + GM-CSF induction therapy and may provide the means for the development of a risk adapted treatment strategy.