Long-Term Survival Analysis in Older Patients with AML Treated Intensively: Positive Impact of Idarubicin in a Cure Fraction Estimation Model.

Background: In older patients with acute myeloid leukemia (AML), current intensive chemotherapy is only associated with a small proportion of long survivors. Despite unsatisfactory overall results with median overall survival (OS) between 6 and 12 months, these long survivors need to be characterized, as eventually cured.

Objective: To address this issue, we performed the present analysis in 884 older patients prospectively treated in two concomitant intensive ALFA trials (median age, 66 years; range, 50 to 85). We used a cure fraction model that provides simultaneous estimates of the proportion of “cured” patients (long survivors) and the distribution of the survival times for uncured patients, as well as prognostic factors separately.

Patients and Methods: Among these 884 patients, 468 with a median age of 60 years were included in the middle age ALFA-9801 trial (Pautas et al. ASH 2007, #162) while 416 with a median age of 72 years were included in the elderly ALFA-9803 trial (Gardin et al. Blood 2007). A frontline randomization between idarubicin (IDA) and daunorubicin (DNR) was included in the two trials, with a total IDA/DNR induction dose of 36/180 mg and 36–48/240 mg in the 9803 and 9801 trial, respectively. After 3+7 like induction, both trials essentially differed by the post-remission chemotherapy, which comprised two intermediate-dose cytarabine (IDAC) cycles in the 9801 trial and a randomization between one standard-dose intensive cycle or six anthracycline-based ambulatory cycles in the 9803 trial. In the 743 patients with available karyotype, cytogenetic distribution was 45 favorable (6%), 556 intermediate (75%), and 142 unfavorable (19%) patients, according to the MRC classification. Statistical analysis was performed with the STATA 10-SE software, using the downloaded strsmix cure fraction model plug in. The following covariates entered the model: age, PS, trial, anthracycline arm, FAB classification, WBC, and cytogenetics. The median follow-up of alive patients was 37 months.

Results: Overall, CR rate was 66% and median OS was 14 months, with estimated 5y OS at 18%. As expected, cytogenetics, PS, WBC and age were identified as of prognostic significance in standard multivariate analysis for both CR achievement and OS. In this multivariate Cox model, anthracycline arm did not significantly impact on OS (P=0.28), even if 5y OS was 24% in the IDA as compared to 11% in the DNR arm. Using the cure fraction model in the whole study population, estimated long-term cure rate was 15% with estimated median OS of 11 months in uncured patients. After adjustment on trial, a higher cure rate was observed in patients with favorable cytogenetics (61% vs 16%, P<0.001) as well as in those treated with IDA (24% vs 13%, P=0.02). IDA appeared to mainly impact on long-term outcome, as median OS of uncured patients was similar in both anthracycline groups (12 vs 13 months, P=0.65). In multivariate analysis, these two factors remained the only significant factors associated with a higher cure rate, with hazard ratio of 6.4 [95% CI, 2.5–16] (P<0.001) for favorable cytogenetics and 3.9 [95% CI, 1.2–13] (P=0.04) for IDA. Cytogenetics, PS, and WBC remained strong independent prognostic factors for OS in uncured patients.

Conclusion: These results indicate that favorable cytogenetics and IDA-based treatment are the two major determinants for a probability of cure in older patients with AML when treated intensively. Given the various post-remission approaches used in ALFA-9801 and 9803 protocols, they also suggest that anthracycline type is of higher importance than the post-remission cytarabine dosing in these patients.