Gene Mutations and Minimal Residual Disease (MRD) As Predictors of Remission Duration in Adults with Core Binding Factor (CBF) Acute Myeloid Leukemia (AML) Treated with High-Dose Cytarabine (HDAC) - First Results of the Prospective French Intergroup CBF-2006 Trial

Abstract

Abstract 410 Background.CBF-AML is a favorable AML subset cytogenetically defined by t(8;21) or inv(16)/t(16;16) rearrangements, respectively responsible for RUNX1-RUNX1T1 (CBFa) or CBFB-MYH11 (CBFb) gene fusion. Nonetheless, relapse incidence may reach 30–40% in these patients. Even if not yet prospectively validated, early MRD monitoring based on chimeric fusion gene transcript quantification by RQ-PCR may help to define patients at higher risk of relapse. Detection of activating KIT and FLT3 gene mutations has also been retrospectively associated with higher relapse rate in these patients. The French CBF-2006 trial (NCT00428558) aimed to prospectively evaluate both markers as predictors of remission duration in homogeneously treated patients. Methods.Patients (18–60y) with newly diagnosed CBF-AML were eligible. All were studied for KIT exon 8/17, FLT3-ITD, FLT3-D835, N-RAS and K-RAS mutations. They were randomized to receive timed-sequential or standard induction. Timed-sequential induction was cytarabine (AraC) 500 mg/m2 CI D1-3 and daunorubicine (DNR) 60 mg/m2 D1-3, followed by AraC 1 g/m2/12h D8-10 and DNR 35 mg/m2 D8-9. Standard arm was AraC 200 mg/m2 CI D1-7 and DNR 60 mg/m2 D1-3, followed by the second sequence at D16-18 only in case of persistent bone marrow (BM) blasts at D15. Patients in first complete remission (CR1) received 3 HDAC consolidation cycles with AraC 3 g/m2/12h D1/3/5. Fusion gene expression was quantified at baseline and before each HDAC cycle (MRD1, MRD2, MRD3) by the 100 × fusion gene/cABL ratio. Patients not reaching a 3-Log ratio reduction at MRD2 were offered allogeneic stem cell transplantation (SCT) in CR1 if they had an HLA-identical donor or could enter a single-agent dasatinib sub-study (see Boissel et al., this meeting). Results.Among the planned 200 patients randomized, 198 met eligibility criteria (96 CBFa and 102 CBFb, compared in Table 1). Overall, incidences of KIT, RAS, FLT3-D835 and FLT3-ITD mutations were 21%, 29%, 10% and 6%, respectively. Patients with KIT and FLT3-D835 mutations had higher WBC (P=0.04 and 0.02) and BM blast percentage (P=0.0004 and 0.05). Those with RAS mutations had lower BM blast percentage (P=0.01) and baseline transcript ratio (P=0.001). Overall, 196 patients reached CR1 and 55 relapsed. At 3 years, probabilities of durable remission, disease-free (DFS) and overall survival from CR were 64%, 61% and 84%, respectively, without differences between the CBF subsets or induction arms. Fifty-five patients did not reach the 3-Log MRD2 reduction (18 CBFa and 37 CBFb, P=0.006), 9 receiving SCT in CR1. Poor MRD2 response correlated with age (P=0.01), WBC (P=0.04), KIT mutations (P=0.04), and initial transcript ratio (P=0.004). Among the characteristics listed in Table 1, shorter remission duration was associated with high WBC (P=0.011), high BM blast percentage (P=0.062), del(9q) (P=0.041), KIT mutation (P=0.048), FLT3-D835 mutation (P=0.077), high initial transcript ratio (P=0.053), and <3-Log MRD2 reduction (P=0.006) in univariate analysis stratified by CBF subtype and treatment arm. In multivariate analysis, the three latter factors, but not KIT mutations, remained independently predictive of shorter remission duration and DFS (P= 0.014, 0.033, 0.002 and 0.001, 0.012, 0.005, respectively). Censoring patients receiving SCT in CR1 yielded similar results. Conclusions.This prospective study of 198 CBF-AML patients homogeneously treated with HDAC revealed significant correlations between initial characteristics and response to therapy. Prognostic analysis did not find that KIT mutations had independent impact on outcome, which was rather influenced by baseline fusion transcript ratio and post-consolidation MRD reduction (reminiscent of Schnittger et al. 2003), and FLT3-D835 mutation. Ongoing pan-genomic analyses might help to further dissect CBF-AML heterogeneity.Table 1Patient characteristicsCBFa (N = 96)CBFb (N = 102)P valuesMedian age42y42y0.37Median WBC, G/L1122<0.001Median BM blast percentage52570.10Therapy-related AML1020.016Loss of Y chromosome*330<0.001Tri 22*018<0.001Tri 8*2150.002Del(9q)*141<0.001Del(7q)/–7*450.99KIT exon 8/17/both mutation**5/16/112/6/00.38FLT3 D835/ITD/both mutation**4/6/015/4/10.11N/K-RAS mutation**13/228/110.001Median baseline transcript ratio30389<0.001>3-Log ratio reduction at MRD2***71580.006*196,**194, and***184 patients studied. Disclosures:No relevant conflicts of interest to declare.