We performed a randomized study of hydroxyurea (HY) versus VP16 in advanced chronic myelomonocytic leukemia (CMML) patients with CMML (according to French-American-British group criteria) and either documented visceral involvement (excluding liver and spleen infiltration) or at least 2 of the following: (1) neutrophils > 16 x 10(9)/I (2) Hemoglobin < 10 g/dL (3) platelets < 100 x 10(9)/L (4) marrow blasts > 5% (5) spleen > 5 cm below costal margin were eligible for this trial. Initial dosage was 1 g/d for HY and 150 mg/week for VP16, orally (doubled in case of visceral involvement). Doses were scheduled to be escalated up to HY 4 g/d and VP16 600 mg/week in the absence of response, and finally adjusted to maintain white blood cells (WBCs) between 5 and 10 x 10(9)/L. Crossing over was scheduled only in case of life threatening visceral involvement or major progression. The major endpoint of the study was survival. The study was closed on first interim analysis that showed a superiority of HY over VP16, after inclusion of 105 pts (HY arm: 53, VP16 arm: 52). Results of the second interim analysis, performed 7 months later, are presented here. Median age was 71 (range 38 to 91), median WBC count 20.10(9)/L (range 10 to 187). Thirteen pts had visceral involvement (3 serous effusions, 8 cutaneous infiltrations, 1 kidney, 1 bone infiltrations). Initial characteristics were similar in the HY and VP16 groups. Median follow up was 11 months in both groups (range 1 to 43+). Response to treatment was seen in 60% of the pts in the HY group, versus 36%, respectively, in the VP16 group (P = .02). Time to response was significantly shorter in the HY group (2.1 v 3.5 months, in the VP16 group, P = .003) and response duration was significantly longer in the HY group (median 24 v 9 months, in the VP16 group, P = .0004). The response rate of patients with visceral involvement was 3 out of 7 in the VP16 arm versus 5 out of 6 in the HY group. Three of the 10 pts crossed over from HY to VP16 responded as compared to 6 pts of the 11 pts crossed over from VP16 to HY. HY yielded better response on leukocytosis (P = .002). The effect on splenomegaly platelets, on hemoglobin level and transfusion requirement was similar in the 2 treatment groups. A significantly higher incidence of alopecia was noted in the VP16 arm (20% v 3%, P = .03). Fourteen (27%) and 20 (38%) patients in the HY and the VP16 group respectively, progressed to acute myeloid leukemia (difference NS). Twenty five (53%) and 44 (83%) patients in the HY and the VP16 group, respectively, had died (P = .002). Median actuarial survival was 20 months in the HY arm, versus 9 months in the VP16 arm (P < 10(-4)). Main factors associated with poor survival were allocation to the VP16 arm, "unfavorable" karyotype (ie, monosomy 7 or complex abnormalities) and anemia. In the HY group, unfavorable karyotype (P = .006), and low hemoglobin level (P = .004) were significantly associated with low response rates. Prognostic factors for poor survival in the HY group were also unfavorable karyotype (P = .001), and low hemoglobin level (P < 10(-4). In conclusion, we found that HY gave higher response rates and better survival than VP16 in advanced CMML. However, even with HY responses were only partial and survival was generally poor. This stresses the need for new agents in the treatment of CMML, that will have to be compared with HY in future randomized studies.