Background: In a previous randomized trial (AML HD98B) of the AMLSG in elderly (>60 yrs) patients with AML (excluding APL), we could demonstrate that all-trans retinoic acid (ATRA) given as adjunct to intensive chemotherapy significantly improved complete remission (CR) rate and overall survival (OS). Our hypothesis for this study was that this beneficial effect of ATRA may be confined to a specific genotypic subset of AML.

Aims: To evaluate the impact of ATRA on clinical outcome in cytogenetic and molecular genetic subsets of AML.

Methods: Between 1998 and 2004, a total of 372 patients were enrolled. 242 patients were randomized for ATRA as adjunct to intensive induction (idarubicin, cytarabine, etoposide) and first consolidation therapy (intermediate-dose cytarabine) (Schlenk et al., Leukemia 2004), followed by a second randomization between further intensive consolidation versus a one-year oral maintenance therapy (Schlenk et al., Leukemia 2006). After an interim analysis in 2003, first randomization was stopped; the following 130 patients received chemotherapy without ATRA. Data from conventional cytogenetics and from fluorescence in-situ hybridization were previously reported (Fröhling et al., Blood 2006). All available leukemia specimens were analyzed for mutations in the genes encoding NPM1, CEBPA, and FLT3.

Results: Median age of the 372 patients was 67 years (range: 61 to 83 yrs); median follow-up time was 68 months. 67% of the patients had de novo AML, 33% had secondary AML. Incidences of mutations were as follows (no. of specimens analyzed): NPM1 mutation, 25% (n=242); FLT3 internal tandem duplication (ITD), 18% (n=263); FLT3 tyrosine kinase domain mutation, 5% (n=244); CEBPA mutation, 8% (CEBPA analyzed only in normal karyotype samples; n=109). Logistic regression analysis performed on all patients identified three significant variables for achievement of CR: the genotype NPM1mut (OR 2.6; 95%-CI, 1.2–5.5), non-adverse karyotype subsuming core-binding factor and cytogenetically normal AML (OR 2.7; 95%-CI, 1.4–4.9); and age (diff. of 10 years, OR 0.58; 95%-CI, 0.3–1.0). Cox proportional hazard model for OS was performed on patients who were up-front randomized for ATRA. This model revealed a significant interaction between ATRA and the NPM1mut/FLT3-ITDneg genotype, confining the beneficial effect of ATRA to patients in that subgroup (HR, 0.29; 95%-CI, 0.10–0.87). Additional variables were log(LDH) (HR, 2.6; 95%-CI, 1.4 – 4.6), age (HR, 1.6; 95%-CI, 1.2– 2.2) and non-adverse karyotype (HR, 0.7; 95%-CI, 0.5–1.0). Univariable analysis revealed a survival after 5 years in patients with the genotype NPM1mut/FLT3-ITDneg of 57% (95%-CI, 28%–78%) in the ATRA-arm (n=16) compared to only 6% (95%-CI, 0%–25%) in the standard-arm (n=14) (p=0.002). In contrast, there was no difference in survival in all other patients with a survival of 2% (95%-CI, 0%–7%) in both arms (p=0.23).

Conclusions: The genotype NPM1mut/FLT3-ITDneg emerges as highly significant predictive factor for response to ATRA in elderly patients with AML. This finding is validated prospectively in the ongoing AMLSG 07–04 study randomizing for ATRA in younger adults (ClinicalTrials.gov Identifier: NCT00151242).

Author notes

Disclosure:Research Funding: Bundesministerium für Bildung und Forschung 01G19981, Pfizer Pharma GmbH.