Long Term Follow-Up of the Gimema GSI 103 AMLE Randomized Trial: Daunoxome Seems To Improve Disease-Free Survival (DFS) of Elderly Patients with Acute Myelogenous Leukemia (AML).

Abstract

The GSI 103 AMLE was a randomized phase III clinical trial to explore the efficacy of DaunoXome (DNX) versus Daunorubicin (DNR) standard treatment in AML elderly patients. From 10/2001 to 2/2004, 301 patients aged 61 to 75 years with a diagnosis of AML according to WHO classification and eligible for intensive chemotherapy were enrolled in this trial by 39 GIMEMA Centers. Patients were randomized to receive induction treatment with DNR (45 mg/sqm day 1 to 3) or DNX (80 mg/sqm day 1 to 3) plus AraC (100 mg/sqm day 1 to 7 by continuous infusion): as consolidation therapy, patients in CR received a further course of the assigned induction treatment. After the consolidation course, patients in CR underwent a 2nd randomization independently from the 1st random, to receive as maintenance therapy either AraC (20 mg twice daily day 1 to 10) plus all-trans retinoic acid (ATRA) (45 mg/sqm day 1 to 10) every 28 days for a maximum of 12 cycles (Arm A) or no further treatment (Arm B). 153 patients were enrolled in the DNR arm and 148 in the DNX arm: the two arms were comparable as to relevant patient characteristics, including karyotype. In the DNR arm, 77 patients (50.3%) achieved CR, 56 (36.6%) were resistant and 20 (13.1%) died during induction (ID). In the DNX arm, 73 patients (49.3%) achieved CR, 47 (31.8%) were resistant and 28 (18.9%) died during induction. Differences between the two arms at univariate analysis were not statistically significant either considering the 3 different types of response or considering ID versus other (Chi-square 0.34 and 0.17 respectively). After CR, the effect of DNX seems to be time-related, due to higher incidence of early deaths in CR (11% vs 3%, p=0.053) and lower incidence of relapse beyond 6 months (at 24 months 60% vs 80%, p=0.064). This translates in a cross of the OS and DFS curves between the two arms, with initial advantage for the DNR arm followed by an advantage for DNX arm after 12 months from diagnosis. A time-dependent covariate Cox model and a landmark analysis on patients in CR after 6 months revealed a better outcome of DNX patients (p= 0.029 and p=0.092, respectively) in the long term DFS follow-up. No difference has been observed in the DFS curves according to maintenance randomization. In conclusion, DNX is as effective as DNR in the induction phase of AML elderly patients, with a trend of increased early toxicity (ID + deaths in CR): however, DNX seems to improve the OS and DFS in the long-term follow-up of these patients, due to a reduction of late relapses. Further trials in more selected elderly patients or in younger patients will be helpful to define the exact role of DNX in AML.